ci: keep browser e2e out of normal pr path

Merge pull request 'ci: reduce scheduled runner load and prep prebaked browsers' (#1628 ) from fix/ci-cron-bots-prebake-1357 into main
ci: reduce scheduled runner load and prep prebaked browsers
2026-05-20 21:02:33 -07:00 · 2026-05-21 03:17:11 +00:00 · 2026-05-20 17:36:09 -07:00 · 2026-05-21 00:12:38 +00:00 · 2026-05-20 23:40:54 +00:00 · 2026-05-20 23:27:58 +00:00
269 changed files with 16758 additions and 6844 deletions
@@ -65,6 +65,11 @@ class ApiError(RuntimeError):
    pass


+class MergePermissionError(ApiError):
+    """Merge failed with a permanent permission error (403/404/405).
+    The queue should skip this PR and move to the next one."""
+
+
@dataclasses.dataclass(frozen=True)
 class MergeDecision:
    ready: bool
@@ -148,15 +153,38 @@ def latest_statuses_by_context(statuses: list[dict]) -> dict[str, dict]:
    return latest


+def _is_tier_low_pending_ok(
+    latest_statuses: dict[str, dict],
+    context: str,
+    pr_labels: set[str],
+) -> bool:
+    """Return True if tier:low PR can tolerate sop-checklist pending state.
+
+    Per sop-checklist-config.yaml tier_failure_mode, tier:low uses soft-fail:
+    sop-checklist posts state=pending when acks are satisfied (missing
+    manager/ceo acks are informational only). The queue should accept
+    pending instead of waiting for success.
+    """
+    if "tier:low" not in pr_labels:
+        return False
+    if "sop-checklist" not in context:
+        return False
+    status = latest_statuses.get(context) or {}
+    return status_state(status) == "pending"
+
+
 def required_contexts_green(
    latest_statuses: dict[str, dict],
    contexts: list[str],
+    pr_labels: set[str] | None = None,
 ) -> tuple[bool, list[str]]:
    missing_or_bad: list[str] = []
    for context in contexts:
        status = latest_statuses.get(context)
        state = status_state(status or {})
        if state != "success":
+            if pr_labels and _is_tier_low_pending_ok(latest_statuses, context, pr_labels):
+                continue  # tier:low soft-fail: accept pending sop-checklist
            missing_or_bad.append(f"{context}={state or 'missing'}")
    return not missing_or_bad, missing_or_bad

@@ -209,6 +237,7 @@ def evaluate_merge_readiness(
    pr_status: dict,
    required_contexts: list[str],
    pr_has_current_base: bool,
+    pr_labels: set[str] | None = None,
 ) -> MergeDecision:
    # Check push-required contexts explicitly instead of combined state.
    # Combined state can be "failure" due to non-blocking jobs
@@ -228,7 +257,7 @@ def evaluate_merge_readiness(
    # The required_contexts list is the authoritative gate — it includes only
    # the checks that actually block merges.
    latest = latest_statuses_by_context(pr_status.get("statuses") or [])
-    ok, missing_or_bad = required_contexts_green(latest, required_contexts)
+    ok, missing_or_bad = required_contexts_green(latest, required_contexts, pr_labels)
    if not ok:
        return MergeDecision(False, "wait", "required contexts not green: " + ", ".join(missing_or_bad))
    return MergeDecision(True, "merge", "ready")
@@ -253,27 +282,32 @@ def get_combined_status(sha: str) -> dict:
    _, combined = api("GET", f"/repos/{OWNER}/{NAME}/commits/{sha}/status")
    if not isinstance(combined, dict):
        raise ApiError(f"status for {sha} response not object")
-    # Fetch full statuses list; 200 covers >99% of real-world runs.
-    # The list is ordered ascending by id (oldest first) — callers must
-    # iterate in reverse to get the newest entry per context.
-    # Best-effort: large repos (main with 550+ statuses) may time out.
-    # On timeout, fall back to the statuses[] already in the combined
-    # response (usually 30 entries — enough for most PRs, enough for
-    # main's early push-required contexts).
+    combined_statuses: list[dict] = combined.get("statuses") or []
    try:
-        _, all_statuses = api(
+        _, all_statuses_raw = api(
            "GET",
            f"/repos/{OWNER}/{NAME}/commits/{sha}/statuses",
            query={"limit": "50"},
        )
-        if isinstance(all_statuses, list):
-            combined["statuses"] = all_statuses
+        if isinstance(all_statuses_raw, list):
+            all_statuses: list[dict] = list(all_statuses_raw)
+        else:
+            all_statuses = []
    except (ApiError, urllib.error.URLError, TimeoutError, OSError) as exc:
-        # URLError covers network-level failures (DNS, refused, timeout).
-        # TimeoutError and OSError cover socket-level timeouts.
        sys.stderr.write(f"::warning::could not fetch full statuses list for {sha[:8]}: {exc}\n")
-        # Fall back to the statuses[] already in the combined response.
-        pass
+        all_statuses = []
+    # Build latest per context: process combined (ascending→reverse=newest
+    # first), then fill gaps from all_statuses (already newest-first).
+    latest: dict[str, dict] = {}
+    for status in reversed(sorted(combined_statuses, key=lambda s: s.get("id") or 0)):
+        ctx = status.get("context")
+        if isinstance(ctx, str) and ctx not in latest:
+            latest[ctx] = status
+    for status in all_statuses:
+        ctx = status.get("context")
+        if isinstance(ctx, str) and ctx not in latest:
+            latest[ctx] = status
+    combined["statuses"] = list(latest.values())
    return combined


@@ -338,7 +372,16 @@ def merge_pull(pr_number: int, *, dry_run: bool) -> None:
    print(f"::notice::merging PR #{pr_number}")
    if dry_run:
        return
-    api("POST", f"/repos/{OWNER}/{NAME}/pulls/{pr_number}/merge", body=payload, expect_json=False)
+    try:
+        api("POST", f"/repos/{OWNER}/{NAME}/pulls/{pr_number}/merge", body=payload, expect_json=False)
+    except ApiError as exc:
+        # Re-raise permission-like errors so process_once can skip this PR.
+        # 403 = no push access, 404 = repo/pr not found, 405 = not allowed.
+        msg = str(exc)
+        for code in ("403", "404", "405"):
+            if code in msg:
+                raise MergePermissionError(msg) from exc
+        raise  # re-raise other ApiErrors unchanged


 def process_once(*, dry_run: bool = False) -> int:
@@ -380,11 +423,13 @@ def process_once(*, dry_run: bool = False) -> int:
    commits = get_pull_commits(pr_number)
    current_base = pr_has_current_base(pr, commits, main_sha)
    pr_status = get_combined_status(head_sha)
+    pr_labels = label_names(pr)
    decision = evaluate_merge_readiness(
        main_status=main_status,
        pr_status=pr_status,
        required_contexts=contexts,
        pr_has_current_base=current_base,
+        pr_labels=pr_labels,
    )

    print(f"::notice::PR #{pr_number} decision={decision.action}: {decision.reason}")
@@ -407,7 +452,25 @@ def process_once(*, dry_run: bool = False) -> int:
                "deferring to next tick"
            )
            return 0
-        merge_pull(pr_number, dry_run=dry_run)
+        try:
+            merge_pull(pr_number, dry_run=dry_run)
+        except MergePermissionError as exc:
+            # Permanent merge failure (HTTP 403/404/405). Post a comment so
+            # maintainers know why, then return 0 so this tick is done.
+            # The PR stays in the queue; future ticks can retry after the
+            # permission issue is resolved.
+            sys.stderr.write(f"::error::merge permission error for PR #{pr_number}: {exc}\n")
+            post_comment(
+                pr_number,
+                (
+                    "merge-queue: merge failed with HTTP 405 'User not allowed to merge PR'. "
+                    "No available token has Can-merge permission on this repo. "
+                    "Fix: grant Can-merge to a token, or add a maintain/admin collaborator. "
+                    "Skipping to next queued PR on next tick."
+                ),
+                dry_run=dry_run,
+            )
+            return 0
        return 0
    return 0

@@ -218,6 +218,31 @@ def is_red(status: dict) -> tuple[bool, list[dict]]:

    `failed_statuses` is the list of per-context entries whose own
    `state` is in the red set; useful for the issue body.
+
+    Cancel-cascade filter (mc#1564, 2026-05-19):
+      Gitea maps BOTH `action_run.status=2 (Failure)` AND
+      `action_run.status=3 (Cancelled)` to commit-status string
+      `"failure"`. On a busy main with
+      `concurrency: cancel-in-progress: true`, every merge burst
+      cancels prior in-flight runs (status=3) — those bubble to the
+      combined-status `failure` and inflate the watchdog's red%,
+      generating phantom `[main-red]` issues (mc#1562/#1552/#1540/...).
+      Canonical Gitea 1.22.6 enum per `models/actions/status.go` +
+      `reference_gitea_action_status_enum_corrected_2026_05_19`:
+          1=Success, 2=Failure, 3=Cancelled, 4=Skipped,
+          5=Waiting, 6=Running, 7=Blocked
+      We only want status=2 (real defects) to file. At the
+      commit-status layer we don't have the integer enum directly
+      (only the `failure` rollup string), so we use the description
+      string Gitea writes when a run is cancelled — empirically
+      `"Has been cancelled"` (verified 2026-05-19 via #1562 body).
+      Real failures show `"Failing after Ns"` and are unaffected.
+      This is option B from mc#1564 (description-string filter, no
+      extra API call). Description-string stability is a soft contract
+      with Gitea; if a future release renames it, the cancel-cascade
+      entries will simply leak back through (visible-not-silent), and
+      we'll either re-pin the string or upgrade to option A (resolve
+      the underlying action_run.status integer via target_url).
    """
    combined = status.get("state")
    statuses = status.get("statuses") or []
@@ -233,11 +258,30 @@ def is_red(status: dict) -> tuple[bool, list[dict]]:
    def _entry_state(s: dict) -> str:
        return s.get("status") or s.get("state") or ""

+    def _is_cancel_cascade(s: dict) -> bool:
+        """status=3 entry per Gitea 1.22.6 description-string contract.
+        Match exactly (after strip) — substring match would catch
+        legitimate test names like "Has been cancelled by the user
+        unexpectedly" in failure logs."""
+        desc = (s.get("description") or "").strip()
+        return desc == "Has been cancelled"
+
    failed = [
        s for s in statuses
-        if isinstance(s, dict) and _entry_state(s) in red_states
+        if isinstance(s, dict)
+        and _entry_state(s) in red_states
+        and not _is_cancel_cascade(s)
    ]
-    return (combined in red_states or bool(failed), failed)
+    # Combined state alone is no longer sufficient — combined=failure
+    # may be 100% cancel-cascade. Drive `red` off the FILTERED list:
+    # if every red-shaped per-entry was cancel-cascade, `failed` is
+    # empty and we report green. Combined-failure with no per-entry
+    # detail (empty `statuses[]`) still trips red — that's the
+    # "CI emitter set combined-status directly" edge case from
+    # render_body's fallback path; we keep filing on it so the
+    # operator sees the breadcrumb.
+    combined_red_no_detail = combined in red_states and not statuses
+    return (bool(failed) or combined_red_no_detail, failed)


 # --------------------------------------------------------------------------
@@ -71,6 +71,12 @@ def build_plan(env: dict[str, str]) -> dict:
        "soak_seconds": _int_env(env, "PROD_AUTO_DEPLOY_SOAK_SECONDS", 60, minimum=0),
        "batch_size": _int_env(env, "PROD_AUTO_DEPLOY_BATCH_SIZE", 3),
        "dry_run": truthy_flag(env.get("PROD_AUTO_DEPLOY_DRY_RUN", "")),
+        # confirm:true ack required by CP /cp/admin/tenants/redeploy-fleet
+        # contract (cp#228 / task #308) for fleet-wide intent. Empty body
+        # / {confirm:false} / {only_slugs:[]} → 400. This caller is the
+        # production auto-deploy step that rolls every live tenant (canary
+        # + fan-out), no slug scoping, so confirm:true is correct.
+        "confirm": True,
    }
    if canary_slug:
        body["canary_slug"] = canary_slug
@@ -100,11 +100,12 @@ printf 'header = "Authorization: token %s"\n' "$GITEA_TOKEN" > "$CURL_AUTH_FILE"
 # (bash trap 'function' EXIT expands variables at trap-fire time, not def time).
 PR_JSON=$(mktemp)
 REVIEWS_JSON=$(mktemp)
+COMMENTS_JSON=$(mktemp)
 TEAM_PROBE_TMP=$(mktemp)
 NA_STATUSES_TMP=""  # declared here so cleanup() always has the var

 cleanup() {
-  rm -f "$CURL_AUTH_FILE" "$PR_JSON" "$REVIEWS_JSON" "$TEAM_PROBE_TMP" "${NA_STATUSES_TMP-}"
+  rm -f "$CURL_AUTH_FILE" "$PR_JSON" "$REVIEWS_JSON" "$COMMENTS_JSON" "$TEAM_PROBE_TMP" "${NA_STATUSES_TMP-}"
 }
 trap cleanup EXIT

@@ -206,7 +207,81 @@ CANDIDATES=$(jq -r --arg author "$PR_AUTHOR" --arg head "$PR_HEAD_SHA" "$JQ_FILT
 debug "candidate non-author approvers: $(echo "$CANDIDATES" | tr '\n' ' ')"

 if [ -z "$CANDIDATES" ]; then
-  echo "::error::${TEAM}-review awaiting non-author APPROVE from ${TEAM} team (no candidates yet)"
+  # --- Guardrail (internal#503): explain the most common false
+  # "no candidates" red. Gitea's review event enum is EXACTLY
+  # APPROVED/REQUEST_CHANGES/COMMENT/PENDING. A wrong value ("APPROVE",
+  # lowercase, ...) is silently accepted (HTTP 200) and stored as
+  # state=PENDING. A correctly-started draft review has an EMPTY body;
+  # a NON-empty body + state==PENDING by a non-author == an intended
+  # verdict mis-filed by a wrong event string. Surface it actionably.
+  # This does NOT change the gate result (still fail-closed below) — it
+  # only converts a mystery red into a named, self-fixing error.
+  MISFILED_FILTER='.[]
+    | select(.state == "PENDING")
+    | select(.dismissed != true)
+    | select(.user.login != $author)
+    | select(((.body // "") | gsub("^\\s+|\\s+$";"") | length) > 0)
+    | "\(.id)\t\(.user.login)"'
+  MISFILED=$(jq -r --arg author "$PR_AUTHOR" "$MISFILED_FILTER" "$REVIEWS_JSON" 2>/dev/null || true)
+  if [ -n "$MISFILED" ]; then
+    echo "::error::${TEAM}-review: non-author review(s) were SUBMITTED but stored as PENDING — almost certainly the wrong Gitea review event string (internal#503)."
+    echo "::error::Gitea accepts ONLY the exact enum APPROVED / REQUEST_CHANGES / COMMENT. 'APPROVE' or lowercase is silently (HTTP 200) filed as PENDING and is invisible to this gate."
+    printf '%s\n' "$MISFILED" | while IFS="$(printf '\t')" read -r _rid _rl; do
+      [ -n "${_rid:-}" ] && echo "::error::  review id=${_rid} by '${_rl}': RE-SUBMIT via POST ${API}/repos/${OWNER}/${NAME}/pulls/${PR_NUMBER}/reviews with {\"event\":\"APPROVED\"} (correct enum) — do NOT edit the DB."
+    done
+  fi
+
+  # --- Fallback (internal#348): check issue comments for agent-approval ---
+  # core-qa-agent and core-security-agent approve via issue comments, NOT
+  # the reviews API. The reviews API returns zero entries for comment-only
+  # approvals. This fallback reads PR issue comments and extracts logins that:
+  #   1. Posted a comment matching the agent-prefix pattern for this gate:
+  #        qa      → "[core-qa-agent] APPROVED"
+  #        security → "[core-security-agent] APPROVED"
+  #      OR posted a generic approval keyword (word-anchored, case-insensitive):
+  #        APPROVED / LGTM / ACCEPTED
+  #   2. Are not the PR author
+  #   3. The team-membership probe below is the authoritative filter.
+  AGENT_PATTERN=""
+  case "$TEAM" in
+    qa)       AGENT_PATTERN="\\[core-qa-agent\\]" ;;
+    security) AGENT_PATTERN="\\[core-security-agent\\]" ;;
+  esac
+  HTTP_CODE=$(curl -sS -o "$COMMENTS_JSON" -w '%{http_code}' \
+    -K "$CURL_AUTH_FILE" "${API}/repos/${OWNER}/${NAME}/issues/${PR_NUMBER}/comments")
+  debug "GET /issues/${PR_NUMBER}/comments → HTTP ${HTTP_CODE}"
+  if [ "$HTTP_CODE" = "200" ]; then
+    # JQ expression: select non-author comments that match either the
+    # agent-prefix pattern (case-insensitive) OR a generic approval keyword.
+    JQ_APPROVALS='
+      .[] |
+      select(.user.login != $author) |
+      . as $cmt |
+      if ($agent_pattern | length) > 0 and ($cmt.body // "" | test($agent_pattern; "i")) then
+        $cmt.user.login
+      elif ($cmt.body // "" | test("\\b(APPROVED|LGTM|ACCEPTED)\\b"; "i")) then
+        $cmt.user.login
+      else
+        empty
+      end
+    '
+    CANDIDATES=$(jq -r \
+      --arg author "$PR_AUTHOR" \
+      --arg agent_pattern "$AGENT_PATTERN" \
+      "$JQ_APPROVALS" \
+      "$COMMENTS_JSON" 2>/dev/null | sort -u)
+    debug "comment-based approval candidates: $(echo "$CANDIDATES" | tr '\n' ' ')"
+
+    if [ -n "$CANDIDATES" ]; then
+      echo "::notice::${TEAM}-review: reviews API found no APPROVED reviews; found $(echo "$CANDIDATES" | wc -w | xargs) comment-based approval candidate(s) — verifying team membership..."
+    fi
+  else
+    debug "could not fetch issue comments (HTTP ${HTTP_CODE})"
+  fi
+fi
+
+if [ -z "${CANDIDATES:-}" ]; then
+  echo "::error::${TEAM}-review awaiting non-author APPROVE from ${TEAM} team (no candidates from reviews API or issue comments)"
  exit 1
 fi

@@ -64,11 +64,41 @@ import argparse
 import json
 import os
 import re
+import resource
 import sys
 import urllib.error
 import urllib.parse
 import urllib.request
-from typing import Any
+from typing import Any, Callable, Iterator
+
+# ---------------------------------------------------------------------------
+# Address-space guardrail (RFC#369 / task #369 follow-up to mc#1242-class OOM).
+#
+# `get_issue_comments` paginates the full comment history of a PR. On
+# bot-relay-heavy PRs (e.g. mc#291, mc#1242) this can balloon past the
+# runner's cgroup memory limit and 137 the job. Cap virtual-address-space
+# at 2 GiB so the script OOMs as a `MemoryError` (catchable / surfaceable)
+# rather than a SIGKILL we can't post a status for.
+#
+# 2 GiB is generous — a 5000-comment PR with 1 KiB minimal-dicts (see
+# get_issue_comments below) fits in ~10 MiB, leaving plenty of headroom
+# for the Python runtime + urllib + json buffers.
+#
+# Skipped under pytest / dry-run where RLIMIT_AS would interfere with
+# test runner memory needs (set SOP_CHECKLIST_NO_RLIMIT=1 to opt out).
+if not os.environ.get("SOP_CHECKLIST_NO_RLIMIT"):
+    try:
+        resource.setrlimit(resource.RLIMIT_AS, (2 * 1024**3, 2 * 1024**3))
+    except (ValueError, OSError):
+        # macOS sometimes refuses RLIMIT_AS; not fatal — the Linux runner
+        # is the only place this matters for the OOM-prevention goal.
+        pass
+
+# Per-comment body cap (task #369). The directive parser walks the body
+# line-by-line looking for ^/sop-ack ^/sop-revoke ^/sop-n/a markers — only
+# the first few KiB matter for that. Cap each comment body so a single
+# pasted-log comment can't push us past the cgroup limit.
+_MAX_BODY_BYTES = int(os.environ.get("SOP_CHECKLIST_MAX_BODY_BYTES") or 8 * 1024)


 # ---------------------------------------------------------------------------
@@ -110,7 +140,7 @@ def normalize_slug(raw: str, numeric_aliases: dict[int, str] | None = None) -> s
 # for /sop-revoke (RFC#351 open question 4 — reason is captured but not
 # yet validated; future iteration may require a min-length).
 _DIRECTIVE_RE = re.compile(
-    r"^[ \t]*/(sop-ack|sop-revoke)[ \t]+([A-Za-z0-9_\- ]+?)(?:[ \t]+(.*))?[ \t]*$",
+    r"^[ \t]*/(sop-ack|sop-revoke|sop-n/a)[ \t]+([A-Za-z0-9_\- ]+?)(?:[ \t]+(.*))?[ \t]*$",
    re.MULTILINE,
 )

@@ -118,19 +148,21 @@ _DIRECTIVE_RE = re.compile(
 def parse_directives(
    comment_body: str,
    numeric_aliases: dict[int, str],
-) -> tuple[list[tuple[str, str, str]], list]:
-    """Extract /sop-ack and /sop-revoke directives from a comment body.
+) -> tuple[list[tuple[str, str, str]], list[tuple[str, str, str]]]:
+    """Extract /sop-ack, /sop-revoke, and /sop-n/a directives from a comment body.

-    Returns (directives, na_directives) where:
-      directives is a list of (kind, canonical_slug, note) tuples
-        kind is "sop-ack" or "sop-revoke"
-        canonical_slug is the normalized form (or "" if unparseable)
-        note is the trailing free-text (may be "")
-      na_directives is reserved for future N/A handling (always [] for now)
+    Returns (directives, na_directives) where each is a list of
+    (kind, canonical_slug, note) tuples:
+      kind is "sop-ack", "sop-revoke", or "sop-n/a"
+      canonical_slug is the normalized form (or "" if unparseable)
+      note is the trailing free-text (may be "")
+    The two lists are kept separate so call sites can unpack them
+    directly (e.g. directives, na_directives = parse_directives(...)).
    """
-    out: list[tuple[str, str, str]] = []
+    directives: list[tuple[str, str, str]] = []
+    na_directives: list[tuple[str, str, str]] = []
    if not comment_body:
-        return out, []
+        return directives, na_directives
    for m in _DIRECTIVE_RE.finditer(comment_body):
        kind = m.group(1)
        raw_slug = (m.group(2) or "").strip()
@@ -160,8 +192,12 @@ def parse_directives(
        note_from_group = (m.group(3) or "").strip()
        # If we collapsed multi-word slug into kebab and there's a
        # trailing-text group too, append it.
-        out.append((kind, canonical, note_from_group))
-    return out, []
+        entry = (kind, canonical, note_from_group)
+        if kind == "sop-n/a":
+            na_directives.append(entry)
+        else:
+            directives.append(entry)
+    return directives, na_directives


 # ---------------------------------------------------------------------------
@@ -174,8 +210,8 @@ def section_marker_present(body: str, marker: str) -> bool:
    on a non-empty line (i.e. the author actually filled it in).

    We require the marker substring AND non-whitespace content on the
-    same line OR within the next line — this prevents trivially-empty
-    checklists like:
+    same line OR within the next non-blank line — this prevents
+    trivially-empty checklists like:

        ## SOP-Checklist
        - [ ] **Comprehensive testing performed**:
@@ -184,9 +220,18 @@ def section_marker_present(body: str, marker: str) -> bool:
    from auto-passing the section-present check. The peer-ack is still
    required, but answering with empty content is captured as a soft
    finding via the section-present test alone.
+
+    NOTE: we scan forward through blank lines (the markdown-header pattern
+    is ## Header\\n\\ncontent) so that a header + blank-line + content
+    structure still satisfies the check. The backward checkbox fallback
+    catches inline markers without a preceding checkbox (mc#1099).
    """
    if not body or not marker:
        return False
+    # Strip trailing whitespace so the blank-line scan below can find
+    # content that appears on the very last line of the body (without
+    # being misled by a trailing \n or spaces).
+    body = body.rstrip()
    body_lower = body.lower()
    marker_lower = marker.lower()
    idx = body_lower.find(marker_lower)
@@ -202,13 +247,44 @@ def section_marker_present(body: str, marker: str) -> bool:
    stripped = re.sub(r"[\s\*:\-\[\]]+", "", line)
    if stripped:
        return True
-    # Fall through: check the NEXT line (multi-line answers).
-    next_line_end = body.find("\n", line_end + 1)
-    if next_line_end < 0:
-        next_line_end = len(body)
-    next_line = body[line_end + 1:next_line_end]
-    stripped_next = re.sub(r"[\s\*:\-\[\]]+", "", next_line)
-    return bool(stripped_next)
+    # Fall through: scan forward, skipping blank-only lines, until we find
+    # non-empty content or run out of body.  Handles:
+    #   ## Header          ← marker line (empty after marker)
+    #                      ← blank line (skipped)
+    #   - actual content   ← found
+    pos = line_end
+    while True:
+        # Skip the current newline and any additional newlines (blank lines).
+        while pos < len(body) and body[pos] == "\n":
+            pos += 1
+        if pos >= len(body):
+            break
+        line_end = body.find("\n", pos)
+        if line_end < 0:
+            line_end = len(body)
+        line = body[pos:line_end]
+        stripped = re.sub(r"[\s\*:\-\[\]]+", "", line)
+        if stripped:
+            return True
+        pos = line_end
+    # Last resort: the marker may appear mid-sentence (e.g.
+    # **Memory/saved-feedback consulted**: No applicable...).
+    # Search backward within the CURRENT LINE only (not preceding lines)
+    # to find a checkbox on the same line before the marker text.
+    # mc#1099 follow-up: memory-consulted detection was failing because
+    # the checkbox was on the same line before the inline marker.
+    _CHECKBOX_RE = re.compile(r"- \[[ x\]]|<input", re.IGNORECASE)
+    line_start = body.rfind("\n", 0, idx) + 1  # 0 if no newline before idx
+    before = body[line_start:idx]
+    m = _CHECKBOX_RE.search(before)
+    if not m:
+        return False
+    # Require meaningful content between the checkbox and the marker text
+    # (markdown formatting like ** or * must also be stripped).
+    # If only whitespace/markdown chars remain, the checkbox line is empty.
+    between = before[m.end() :]
+    stripped_between = re.sub(r"[\s\*:#\[\]_\-]+", "", between)
+    return bool(stripped_between)


 # ---------------------------------------------------------------------------
@@ -222,6 +298,7 @@ def compute_ack_state(
    items_by_slug: dict[str, dict[str, Any]],
    numeric_aliases: dict[int, str],
    team_membership_probe: "callable[[str, list[str]], list[str]]",
+    high_risk: bool = False,
 ) -> dict[str, dict[str, Any]]:
    """Compute per-item ack state.

@@ -251,8 +328,7 @@ def compute_ack_state(
        user = (c.get("user") or {}).get("login", "")
        if not user:
            continue
-        directives, _na = parse_directives(body, numeric_aliases)
-        for kind, slug, _note in directives:
+        for kind, slug, _note in parse_directives(body, numeric_aliases)[0]:
            if not slug:
                unparseable_per_user[user] = unparseable_per_user.get(user, 0) + 1
                continue
@@ -285,11 +361,16 @@ def compute_ack_state(
    for slug, candidates in pending_team_check.items():
        if not candidates:
            continue
-        required = items_by_slug[slug]["required_teams"]
+        # Risk-class-aware required-teams resolution (RFC#450 Option C):
+        # high-risk PRs use `required_teams_high_risk` (when set on the
+        # item); default class uses `required_teams`. The probe closure
+        # is built with the same high_risk flag so the two reads are
+        # always consistent (both sites share `resolve_required_teams`).
+        required = resolve_required_teams(items_by_slug[slug], high_risk)
        approved = team_membership_probe(slug, candidates)  # returns subset
        rejected_not_in_team[slug] = [u for u in candidates if u not in approved]
        ackers_per_slug[slug] = approved
-        # Stash required teams for description rendering.
+        # Stash resolved teams for description rendering.
        items_by_slug[slug]["_required_resolved"] = required

    return {
@@ -304,6 +385,63 @@ def compute_ack_state(
    }


+# ---------------------------------------------------------------------------
+# N/A-gate evaluation
+# ---------------------------------------------------------------------------
+
+
+def compute_na_state(
+    comments: list[dict[str, Any]],
+    author: str,
+    na_gates: dict[str, Any],
+    probe: Callable[[str, list[str]], list[str]],
+) -> dict[str, dict[str, Any]]:
+    """Evaluate which N/A gates have a valid declaration from a team member.
+
+    Returns dict[gate_name, dict] where each dict has:
+      declared: bool — at least one valid non-author team-member declared N/A
+      decl_ackers: list[str] — usernames who declared this gate N/A
+      rejected: dict with keys:
+        not_in_team: list[str] — users who tried but aren't in required teams
+    """
+    # Build per-user latest N/A directive (most-recent wins per RFC#324).
+    latest_na: dict[str, tuple[str, str]] = {}  # user → (gate, note)
+    for c in comments:
+        body = c.get("body", "") or ""
+        user = (c.get("user") or {}).get("login", "")
+        if not user:
+            continue
+        for kind, gate, note in parse_directives(body, {})[1]:
+            # [1] = na_directives only
+            if gate in na_gates:
+                latest_na[user] = (gate, note)
+
+    result: dict[str, dict[str, Any]] = {}
+    for gate, gate_cfg in na_gates.items():
+        result[gate] = {
+            "declared": False,
+            "decl_ackers": [],
+            "rejected": {"not_in_team": []},
+        }
+        decl_ackers: list[str] = []
+        not_in_team: list[str] = []
+        for user, (g, _note) in latest_na.items():
+            if g != gate:
+                continue
+            if user == author:
+                continue  # authors cannot self-declare N/A
+            approved = probe(gate, [user])
+            if approved:
+                decl_ackers.append(user)
+            else:
+                not_in_team.append(user)
+        result[gate]["declared"] = bool(decl_ackers)
+        result[gate]["decl_ackers"] = decl_ackers
+        result[gate]["rejected"]["not_in_team"] = not_in_team
+
+    return result
+
+
 # ---------------------------------------------------------------------------
 # Gitea API client
 # ---------------------------------------------------------------------------
@@ -352,16 +490,35 @@ class GiteaClient:
            raise RuntimeError(f"GET pulls/{pr} → HTTP {code}: {data!r}")
        return data

-    def get_issue_comments(
-        self, owner: str, repo: str, issue: int
-    ) -> list[dict[str, Any]]:
-        # Paginate. Gitea default page size 50.
-        out: list[dict[str, Any]] = []
+    def iter_issue_comments(
+        self, owner: str, repo: str, issue: int, page_size: int = 50
+    ) -> Iterator[dict[str, Any]]:
+        """Stream comments page-by-page, yielding ONE minimal-dict per comment.
+
+        Each yielded comment carries ONLY the fields the gate actually reads
+        — `{"user": {"login": str}, "body": str}` — and DROPS the much
+        larger Gitea-API extras (html_url, pull_request_url, issue_url,
+        assets, created_at, updated_at, id, original_author_*).
+
+        Memory motivation (task #369 / mc#1242-class OOM): full Gitea
+        comment dicts are ~2 KiB median + ~3 KiB p95. On PRs with several
+        thousand bot-relay comments the eager `list[full_dict]` shape used
+        previously pushed runner anon-rss past the cgroup limit. The
+        minimal-dict shape is ~10-20x smaller (typically ~50-100B Python
+        overhead + the body string).
+
+        The two downstream consumers (`compute_ack_state`,
+        `compute_na_state`) each iterate the comment list exactly once and
+        read only `body` + `user.login`, so dropping every other field is
+        safe. They still receive `list[dict[str, Any]]`-shaped objects so
+        the test fixtures (which already used the minimal shape) keep
+        working with no fixture changes.
+        """
        page = 1
        while True:
            code, data = self._req(
                "GET",
-                f"/repos/{owner}/{repo}/issues/{issue}/comments?limit=50&page={page}",
+                f"/repos/{owner}/{repo}/issues/{issue}/comments?limit={page_size}&page={page}",
            )
            if code != 200:
                raise RuntimeError(
@@ -369,10 +526,41 @@ class GiteaClient:
                )
            if not data:
                break
-            out.extend(data)
-            if len(data) < 50:
+            for c in data:
+                # Minimal projection — drop ALL fields the gate doesn't read.
+                user_login = ((c.get("user") or {}).get("login") or "") if isinstance(c, dict) else ""
+                body = (c.get("body") if isinstance(c, dict) else "") or ""
+                # Body-size guardrail: huge comments (e.g. pasted CI logs) can
+                # individually be MiBs. The directive parser only needs the
+                # first ~8 KiB to find /sop-ack /sop-revoke /sop-n/a markers
+                # — anything past that is filler. Truncate at 8 KiB so a
+                # single oversized comment can't OOM the runner.
+                if len(body) > _MAX_BODY_BYTES:
+                    body = body[:_MAX_BODY_BYTES]
+                yield {"user": {"login": user_login}, "body": body}
+            if len(data) < page_size:
                break
            page += 1
+
+    def get_issue_comments(
+        self,
+        owner: str,
+        repo: str,
+        issue: int,
+        max_comments: int | None = None,
+    ) -> list[dict[str, Any]]:
+        """Paginate + collect minimal comment dicts. See `iter_issue_comments`
+        for the per-comment shape and the OOM-prevention rationale.
+
+        `max_comments` (optional, default unbounded): hard cap. When the cap
+        is hit we stop fetching further pages and the caller surfaces a
+        soft 'skipping due to volume' status (see main()).
+        """
+        out: list[dict[str, Any]] = []
+        for c in self.iter_issue_comments(owner, repo, issue):
+            out.append(c)
+            if max_comments is not None and len(out) >= max_comments:
+                break
        return out

    def resolve_team_id(self, org: str, team_name: str) -> int | None:
@@ -663,6 +851,42 @@ def get_tier_mode(pr: dict[str, Any], cfg: dict[str, Any]) -> str:
    return default_mode


+def is_high_risk(pr: dict[str, Any], cfg: dict[str, Any]) -> bool:
+    """Return True when the PR is high-risk per RFC#450 Option C.
+
+    A PR is high-risk when ANY of:
+      - it carries the `tier:high` label (mechanically strictest tier), or
+      - it carries any label listed in cfg.high_risk_labels.
+
+    High-risk PRs use `required_teams_high_risk` (when set on an item)
+    instead of the default `required_teams`. Items without
+    `required_teams_high_risk` are unaffected (the default applies).
+
+    Governance fix for internal#442 — closes the inconsistency between
+    sop-tier-check (tier-aware) and sop-checklist (was tier-blind).
+    """
+    label_set = {(l.get("name") or "") for l in (pr.get("labels") or [])}
+    if "tier:high" in label_set:
+        return True
+    high_risk_labels = set(cfg.get("high_risk_labels") or [])
+    return bool(label_set & high_risk_labels)
+
+
+def resolve_required_teams(item: dict[str, Any], high_risk: bool) -> list[str]:
+    """Pick the active required_teams list for an item.
+
+    When high_risk is True AND the item declares a non-empty
+    `required_teams_high_risk`, return that. Else fall back to
+    `required_teams`. Keeping this in one helper means the gate's
+    decision shape stays single-sited even as items grow.
+    """
+    if high_risk:
+        elevated = item.get("required_teams_high_risk") or []
+        if elevated:
+            return list(elevated)
+    return list(item.get("required_teams") or [])
+
+
 def main(argv: list[str] | None = None) -> int:
    p = argparse.ArgumentParser()
    p.add_argument("--owner", required=True)
@@ -688,6 +912,17 @@ def main(argv: list[str] | None = None) -> int:
            "thing BP sees is the POSTed status. Useful for local debugging."
        ),
    )
+    p.add_argument(
+        "--max-comments",
+        type=int,
+        default=int(os.environ.get("SOP_CHECKLIST_MAX_COMMENTS") or 5000),
+        help=(
+            "Hard cap on comments fetched from the PR. Above this we post "
+            "a SOFT-pending status with a 'skipping due to volume' note "
+            "instead of OOM'ing the runner (task #369). Override with the "
+            "SOP_CHECKLIST_MAX_COMMENTS env var. Set 0 to disable the cap."
+        ),
+    )
    args = p.parse_args(argv)

    token = os.environ.get("GITEA_TOKEN", "")
@@ -698,6 +933,7 @@ def main(argv: list[str] | None = None) -> int:
    cfg = load_config(args.config)
    items: list[dict[str, Any]] = cfg["items"]
    items_by_slug = {it["slug"]: it for it in items}
+    na_gates: dict[str, Any] = cfg.get("n/a_gates", {})
    numeric_aliases = {
        int(it["numeric_alias"]): it["slug"] for it in items if it.get("numeric_alias")
    }
@@ -720,7 +956,24 @@ def main(argv: list[str] | None = None) -> int:
        print("::error::PR payload missing user.login or head.sha", file=sys.stderr)
        return 1

-    comments = client.get_issue_comments(args.owner, args.repo, args.pr)
+    max_comments_cap = args.max_comments if args.max_comments and args.max_comments > 0 else None
+    comments = client.get_issue_comments(
+        args.owner, args.repo, args.pr, max_comments=max_comments_cap
+    )
+
+    # Volume short-circuit: PRs with thousands of bot-relay comments
+    # (the mc#1242-class OOM source) get a soft 'volume-skipped' status
+    # so the gate doesn't churn the runner; reviewers can re-trigger by
+    # editing the PR or filing a fresh PR with the housekeeping comments
+    # split off. Cap-hit means we couldn't see the WHOLE history, so we
+    # can't fairly post failure — pending is the safe default.
+    volume_skipped = bool(max_comments_cap and len(comments) >= max_comments_cap)
+
+    # High-risk classification (RFC#450 Option C, governance fix for
+    # internal#442). Computed ONCE per PR — used by both the probe
+    # closure and compute_ack_state so the elevation decision is
+    # single-sited.
+    high_risk = is_high_risk(pr, cfg)

    # Build team-membership probe closure that caches results per
    # (user, team-id) so a user acking multiple items only triggers
@@ -728,8 +981,34 @@ def main(argv: list[str] | None = None) -> int:
    team_member_cache: dict[tuple[str, int], bool | None] = {}

    def probe(slug: str, users: list[str]) -> list[str]:
-        item = items_by_slug[slug]
-        team_names: list[str] = item["required_teams"]
+        # `slug` may be either an items-key (compute_ack_state caller) OR
+        # an n/a-gate key (compute_na_state caller). Previously this hard
+        # KeyError'd on the n/a-gate path when slug was e.g. "security-review"
+        # — that's a config gate, not an item — so the gate would crash
+        # instead of falling back to the gate's own required_teams. Fix
+        # task #369 follow-up to issue #355.
+        if slug in items_by_slug:
+            item = items_by_slug[slug]
+            team_names: list[str] = resolve_required_teams(item, high_risk)
+        elif slug in na_gates:
+            # n/a-gate configs carry `required_teams` directly (see
+            # sop-checklist-config.yaml: n/a_gates.<gate>.required_teams).
+            gate_cfg = na_gates[slug] or {}
+            team_names = list(gate_cfg.get("required_teams") or [])
+            if not team_names:
+                print(
+                    f"::warning::n/a-gate '{slug}' has no required_teams; "
+                    "fail-closed (no users will be approved)",
+                    file=sys.stderr,
+                )
+        else:
+            # Unknown slug — fail closed, log so we can find config drift.
+            print(
+                f"::warning::probe() called with slug '{slug}' which is "
+                f"neither an items entry nor an n/a-gate; fail-closed",
+                file=sys.stderr,
+            )
+            return []
        # Resolve names → ids. NOTE: orgs/{org}/teams/search may not be
        # available — fall back to the list endpoint.
        team_ids: list[int] = []
@@ -774,7 +1053,9 @@ def main(argv: list[str] | None = None) -> int:
                    # may still find membership in another team.
        return approved

-    ack_state = compute_ack_state(comments, author, items_by_slug, numeric_aliases, probe)
+    ack_state = compute_ack_state(
+        comments, author, items_by_slug, numeric_aliases, probe, high_risk=high_risk
+    )
    body_state = {it["slug"]: section_marker_present(body, it["pr_section_marker"]) for it in items}

    state, description = render_status(items, ack_state, body_state)
@@ -785,9 +1066,21 @@ def main(argv: list[str] | None = None) -> int:
        # were not required (vs a tier:medium+ PR that truly passed all acks).
        state = "success"
        description = f"[info tier:low] {description}"
+    if volume_skipped:
+        # Above the comment-cap — we may have a partial view. Soft-pend
+        # so neither BP nor the author gets stuck; surface the cap so
+        # reviewers know what's up. No-block at the gate level.
+        state = "pending"
+        description = (
+            f"[volume-skipped] comment-cap={max_comments_cap} hit; please file "
+            f"a fresh PR with bot-relay history split off (#369). {description}"
+        )

    # Diagnostics to job log.
-    print(f"::notice::PR #{args.pr} author={author} head={head_sha[:7]} mode={mode}")
+    print(
+        f"::notice::PR #{args.pr} author={author} head={head_sha[:7]} "
+        f"mode={mode} risk_class={'high' if high_risk else 'default'}"
+    )
    for it in items:
        slug = it["slug"]
        ackers = ack_state[slug]["ackers"]
@@ -818,6 +1111,46 @@ def main(argv: list[str] | None = None) -> int:
        description=description, target_url=target_url,
    )
    print(f"::notice::status posted: {args.status_context} → {state}")
+
+    # --- N/A gate status (RFC#324 §N/A follow-up) ---
+    # Post a separate status so review-check.sh can discover N/A declarations
+    # and waive the Gitea-approve requirement for that gate.
+    na_state: dict[str, dict[str, Any]] = {}
+    if na_gates:
+        na_state = compute_na_state(comments, author, na_gates, probe)
+
+        na_descs: list[str] = []
+        for gate, s in na_state.items():
+            if s["declared"]:
+                na_descs.append(gate)
+            decl = s["decl_ackers"]
+            rej = s["rejected"]["not_in_team"]
+            if decl:
+                print(f"::notice::  [N/A OK] {gate} — declared by {','.join(decl)}")
+            if rej:
+                print(
+                    f"::notice::  [N/A REJ] {gate} — not-in-team: {','.join(rej)}",
+                    file=sys.stderr,
+                )
+
+        na_desc = ", ".join(sorted(na_descs)) if na_descs else "(none)"
+        na_status_state = "success" if na_descs else "pending"
+        # review-check.sh reads the description to discover which gates are N/A.
+        # Include the gate names so it can grep for them.
+        na_description = f"N/A: {na_desc}" if na_descs else "N/A: (none)"
+
+        if not args.dry_run:
+            client.post_status(
+                args.owner, args.repo, head_sha,
+                state=na_status_state,
+                context="sop-checklist / na-declarations (pull_request)",
+                description=na_description,
+                target_url=target_url,
+            )
+            print(
+                f"::notice::na-declarations status → {na_status_state}: {na_description}"
+            )
+
    # By default exit 0 — the POSTed status IS the gate, NOT the job
    # conclusion. If the job exits 1 BP will see TWO failure signals
    # (one from the job's auto-status, one from our POST), making the
@@ -17,6 +17,9 @@ Scenarios:
  T8_team_not_member          — team membership → 404 (not a member) → exit 1
  T9_team_403                — team membership → 403 (token not in team) → exit 1
  T14_non_default_base        — open PR targeting staging → script exits 0 (no-op)
+  T15_comments_agent_approval — reviews empty; comments have "[core-qa-agent] APPROVED" → exit 0
+  T16_comments_generic_approval — reviews empty; comments have "APPROVED" by team member → exit 0
+  T17_comments_no_approval   — reviews empty; comments have no approval keywords → exit 1

 Usage:
  FIXTURE_STATE_DIR=/tmp/x python3 _review_check_fixture.py 8080
@@ -97,7 +100,9 @@ class Handler(http.server.BaseHTTPRequestHandler):
        # GET /repos/{owner}/{name}/pulls/{pr_number}/reviews
        m = re.match(r"^/api/v1/repos/([^/]+)/([^/]+)/pulls/(\d+)/reviews$", path)
        if m:
-            if sc in ("T4_reviews_empty", "T5_reviews_only_author"):
+            if sc in ("T4_reviews_empty", "T5_reviews_only_author",
+                      "T15_comments_agent_approval", "T16_comments_generic_approval",
+                      "T17_comments_no_approval"):
                return self._json(200, [])
            if sc == "T6_reviews_dismissed":
                return self._json(200, [{
@@ -116,6 +121,28 @@ class Handler(http.server.BaseHTTPRequestHandler):
                {"state": "APPROVED", "dismissed": False, "user": {"login": "core-devops"}, "commit_id": "abc1234"},
            ])

+        # GET /repos/{owner}/{name}/issues/{pr_number}/comments
+        m = re.match(r"^/api/v1/repos/([^/]+)/([^/]+)/issues/(\d+)/comments$", path)
+        if m:
+            if sc == "T15_comments_agent_approval":
+                return self._json(200, [
+                    {"user": {"login": "core-qa-agent"}, "body": "[core-qa-agent] APPROVED this PR. Good changes.", "id": 1},
+                    {"user": {"login": "alice"}, "body": "I authored this PR", "id": 2},
+                    {"user": {"login": "random-user"}, "body": "Looks okay to me", "id": 3},
+                ])
+            if sc == "T16_comments_generic_approval":
+                return self._json(200, [
+                    {"user": {"login": "core-qa-agent"}, "body": "APPROVED — all acceptance criteria met", "id": 1},
+                    {"user": {"login": "alice"}, "body": "-authored", "id": 2},
+                ])
+            if sc == "T17_comments_no_approval":
+                return self._json(200, [
+                    {"user": {"login": "alice"}, "body": "I authored this PR", "id": 1},
+                    {"user": {"login": "random-user"}, "body": "Looks okay to me", "id": 2},
+                ])
+            # Default scenarios (T1–T9, T14): no comments
+            return self._json(200, [])
+
        # GET /teams/{team_id}/members/{username}
        m = re.match(r"^/api/v1/teams/(\d+)/members/([^/]+)$", path)
        if m:
@@ -127,6 +154,12 @@ class Handler(http.server.BaseHTTPRequestHandler):
            # T7_team_member: member
            return self._empty(204)

+        # GET /repos/{owner}/{name}/statuses/{sha} — for N/A declaration check
+        m = re.match(r"^/api/v1/repos/([^/]+)/([^/]+)/statuses/([a-f0-9]+)$", path)
+        if m:
+            # All comment-based scenarios have no N/A declarations
+            return self._json(200, [])
+
        return self._json(404, {"path": path, "msg": "fixture: no route"})

    def do_POST(self):
@@ -118,3 +118,13 @@ def test_merge_decision_updates_stale_pr_before_merge():

    assert decision.ready is False
    assert decision.action == "update"
+
+
+def test_MergePermissionError_inherits_from_ApiError():
+    assert issubclass(mq.MergePermissionError, mq.ApiError)
+
+
+def test_MergePermissionError_message_preserved():
+    exc = mq.MergePermissionError("POST /merge -> HTTP 405: User not allowed")
+    assert "405" in str(exc)
+    assert "User not allowed" in str(exc)
@@ -36,9 +36,37 @@ def test_build_plan_defaults_to_staging_sha_target_and_prod_cp():
        "soak_seconds": 60,
        "batch_size": 3,
        "dry_run": False,
+        # cp#228 / task #308: fleet-wide intent must carry confirm:true.
+        "confirm": True,
    }


+def test_build_plan_always_sets_confirm_true_for_fleet_intent():
+    """Regression guard: every plan body MUST carry confirm:true.
+
+    CP /cp/admin/tenants/redeploy-fleet (cp#228) returns 400 on empty
+    body / {confirm:false} / {only_slugs:[]} to prevent accidental
+    fleet-wide mutation. This caller is fleet-wide intent (canary +
+    fan-out, no slug scoping), so the plan MUST carry confirm:true.
+    Pairs with cp#228's TestRedeployFleet_EmptyBodyReturns400 +
+    TestRedeployFleet_ConfirmTrueProceeds.
+    """
+    plan = prod.build_plan({"GITHUB_SHA": "abcdef1234567890"})
+    assert plan["body"]["confirm"] is True
+
+    # Operator-overridable knobs do NOT drop the ack.
+    plan = prod.build_plan(
+        {
+            "GITHUB_SHA": "abcdef1234567890",
+            "PROD_AUTO_DEPLOY_SOAK_SECONDS": "0",
+            "PROD_AUTO_DEPLOY_BATCH_SIZE": "10",
+            "PROD_AUTO_DEPLOY_DRY_RUN": "true",
+            "PROD_AUTO_DEPLOY_CANARY_SLUG": "",
+        }
+    )
+    assert plan["body"]["confirm"] is True
+
+
 def test_build_plan_rejects_non_prod_cp_without_explicit_override():
    try:
        prod.build_plan(
@@ -334,6 +334,31 @@ assert_contains "T12 jq: core-devops (non-author APPROVED) in candidates" "core-
 assert_eq "T12 jq: alice (author) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^alice$' || true)"
 assert_eq "T12 jq: carol (dismissed) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^carol$' || true)"

+# T15 — comment-based approval via agent prefix pattern → exit 0
+echo
+echo "== T15 comment agent-prefix approval =="
+T15_OUT=$(run_review_check "T15_comments_agent_approval")
+T15_RC=$(cat "$FIX_STATE_DIR/last_rc")
+assert_eq "T15 exit code 0 (agent-comment approval + team member)" "0" "$T15_RC"
+assert_contains "T15 comment fallback notice" "comment-based approval" "$T15_OUT"
+assert_contains "T15 core-qa-agent APPROVED" "APPROVED by core-qa-agent" "$T15_OUT"
+
+# T16 — comment-based approval via generic APPROVED keyword → exit 0
+echo
+echo "== T16 comment generic keyword approval =="
+T16_OUT=$(run_review_check "T16_comments_generic_approval")
+T16_RC=$(cat "$FIX_STATE_DIR/last_rc")
+assert_eq "T16 exit code 0 (generic-approval comment + team member)" "0" "$T16_RC"
+assert_contains "T16 comment fallback notice" "comment-based approval" "$T16_OUT"
+
+# T17 — no approval keywords in comments → exit 1
+echo
+echo "== T17 comments with no approval keywords =="
+T17_OUT=$(run_review_check "T17_comments_no_approval")
+T17_RC=$(cat "$FIX_STATE_DIR/last_rc")
+assert_eq "T17 exit code 1 (no candidates from comments)" "1" "$T17_RC"
+assert_contains "T17 no candidates error" "no candidates from reviews API or issue comments" "$T17_OUT"
+
 echo
 echo "------"
 echo "PASS=$PASS FAIL=$FAIL"
@@ -551,3 +551,456 @@ class TestEndToEndAckFlow(unittest.TestCase):

 if __name__ == "__main__":
    unittest.main(verbosity=2)
+
+
+# ---------------------------------------------------------------------------
+# compute_na_state
+# ---------------------------------------------------------------------------
+
+
+class TestComputeNaState(unittest.TestCase):
+    """Tests for /sop-n/a directive evaluation."""
+
+    def test_no_na_declarations(self):
+        cfg = sop.load_config(CONFIG_PATH)
+        na_gates = cfg.get("n/a_gates", {})
+        comments = []
+        na_state = sop.compute_na_state(comments, "alice", na_gates, lambda *_: [])
+        self.assertFalse(na_state["qa-review"]["declared"])
+        self.assertFalse(na_state["security-review"]["declared"])
+
+    def test_na_declared_by_authorized_user(self):
+        cfg = sop.load_config(CONFIG_PATH)
+        na_gates = cfg.get("n/a_gates", {})
+        comments = [_comment("bob", "/sop-n/a qa-review N/A: pure tooling change")]
+        na_state = sop.compute_na_state(comments, "alice", na_gates, lambda g, u: u)
+        self.assertTrue(na_state["qa-review"]["declared"])
+        self.assertEqual(na_state["qa-review"]["decl_ackers"], ["bob"])
+
+    def test_na_declared_by_unauthorized_user_rejected(self):
+        cfg = sop.load_config(CONFIG_PATH)
+        na_gates = cfg.get("n/a_gates", {})
+        comments = [_comment("mallory", "/sop-n/a qa-review N/A: not real team")]
+        na_state = sop.compute_na_state(comments, "alice", na_gates, lambda g, u: [])
+        self.assertFalse(na_state["qa-review"]["declared"])
+        self.assertEqual(na_state["qa-review"]["rejected"]["not_in_team"], ["mallory"])
+
+    def test_author_cannot_self_declare_na(self):
+        cfg = sop.load_config(CONFIG_PATH)
+        na_gates = cfg.get("n/a_gates", {})
+        comments = [_comment("alice", "/sop-n/a qa-review N/A: I am the author")]
+        na_state = sop.compute_na_state(comments, "alice", na_gates, lambda g, u: u)
+        self.assertFalse(na_state["qa-review"]["declared"])
+
+    def test_parse_directives_separates_na_from_ack(self):
+        directives, na_directives = sop.parse_directives(
+            "/sop-ack comprehensive-testing\n/sop-n/a qa-review N/A: no surface",
+            {},
+        )
+        self.assertEqual(len(directives), 1)
+        self.assertEqual(directives[0][0], "sop-ack")
+        self.assertEqual(len(na_directives), 1)
+        self.assertEqual(na_directives[0][0], "sop-n/a")
+        self.assertEqual(na_directives[0][1], "qa-review")
+
+
+# ---------------------------------------------------------------------------
+# RFC#450 Option C — risk-classed two-eyes (governance fix for internal#442)
+# ---------------------------------------------------------------------------
+
+
+class TestIsHighRisk(unittest.TestCase):
+    """The high-risk predicate decides which required_teams list applies.
+
+    Predicate: tier:high label OR any label in cfg.high_risk_labels.
+    """
+
+    def setUp(self):
+        self.cfg = sop.load_config(CONFIG_PATH)
+
+    def test_no_labels_is_default_class(self):
+        pr = {"labels": []}
+        self.assertFalse(sop.is_high_risk(pr, self.cfg))
+
+    def test_tier_high_is_high_risk(self):
+        pr = {"labels": [{"name": "tier:high"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_tier_low_is_default_class(self):
+        pr = {"labels": [{"name": "tier:low"}]}
+        self.assertFalse(sop.is_high_risk(pr, self.cfg))
+
+    def test_tier_medium_is_default_class(self):
+        # tier:medium alone is NOT high-risk (Option C — medium routes
+        # to the wider engineers OR-set).
+        pr = {"labels": [{"name": "tier:medium"}]}
+        self.assertFalse(sop.is_high_risk(pr, self.cfg))
+
+    def test_area_security_label_is_high_risk(self):
+        pr = {"labels": [{"name": "tier:medium"}, {"name": "area:security"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_area_schema_label_is_high_risk(self):
+        pr = {"labels": [{"name": "area:schema"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_area_identity_label_is_high_risk(self):
+        pr = {"labels": [{"name": "area:identity"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_area_fleet_image_label_is_high_risk(self):
+        pr = {"labels": [{"name": "area:fleet-image"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_area_gate_meta_label_is_high_risk(self):
+        # Gate-meta = changes to sop-checklist/sop-tier-check itself.
+        pr = {"labels": [{"name": "area:gate-meta"}]}
+        self.assertTrue(sop.is_high_risk(pr, self.cfg))
+
+    def test_unknown_area_label_is_default_class(self):
+        pr = {"labels": [{"name": "area:docs"}]}
+        self.assertFalse(sop.is_high_risk(pr, self.cfg))
+
+
+class TestResolveRequiredTeams(unittest.TestCase):
+    """The team resolver picks the elevated list only for high-risk PRs
+    AND only when the item declares one — items without an elevated
+    list always use the default required_teams."""
+
+    def test_default_class_uses_default_teams(self):
+        item = {"required_teams": ["engineers", "managers", "ceo"], "required_teams_high_risk": ["ceo"]}
+        self.assertEqual(
+            sop.resolve_required_teams(item, high_risk=False),
+            ["engineers", "managers", "ceo"],
+        )
+
+    def test_high_risk_uses_elevated_teams(self):
+        item = {"required_teams": ["engineers", "managers", "ceo"], "required_teams_high_risk": ["ceo"]}
+        self.assertEqual(
+            sop.resolve_required_teams(item, high_risk=True),
+            ["ceo"],
+        )
+
+    def test_high_risk_without_elevated_falls_back_to_default(self):
+        # Items that don't declare required_teams_high_risk (e.g.
+        # comprehensive-testing, staging-smoke) are unaffected by risk-class.
+        item = {"required_teams": ["engineers"]}
+        self.assertEqual(
+            sop.resolve_required_teams(item, high_risk=True),
+            ["engineers"],
+        )
+
+    def test_empty_elevated_list_falls_back_to_default(self):
+        # A defensive case: required_teams_high_risk: [] should not
+        # silently lock out all approvers — fall back to the default
+        # so the gate stays satisfiable. (Tightening should remove the
+        # key, not set it to empty.)
+        item = {"required_teams": ["engineers"], "required_teams_high_risk": []}
+        self.assertEqual(
+            sop.resolve_required_teams(item, high_risk=True),
+            ["engineers"],
+        )
+
+
+class TestRootCauseAckEligibilityWidened(unittest.TestCase):
+    """Closes internal#442: a non-author engineers-team ack now satisfies
+    root-cause / no-backwards-compat for the default class.
+
+    The dead-managers/ceo-persona-token gridlock is the symptom; the
+    root cause is that sop-checklist ignored tier-class. These tests
+    pin the new wider-default behavior so it can't regress silently.
+    """
+
+    def setUp(self):
+        self.items = _items_by_slug()
+        self.aliases = _numeric_aliases()
+
+    @staticmethod
+    def _approve_only(allowed):
+        return lambda slug, users: [u for u in users if u in allowed]
+
+    def test_engineers_ack_satisfies_root_cause_default_class(self):
+        # Bob is in engineers only (not managers, not ceo). Default class.
+        comments = [_comment("bob", "/sop-ack root-cause")]
+        # Probe: bob is approved because root-cause now lists engineers.
+        probe = self._approve_only({"bob"})
+        state = sop.compute_ack_state(
+            comments, "alice", self.items, self.aliases, probe, high_risk=False
+        )
+        self.assertEqual(state["root-cause"]["ackers"], ["bob"])
+
+    def test_engineers_ack_satisfies_no_backwards_compat_default_class(self):
+        comments = [_comment("bob", "/sop-ack no-backwards-compat")]
+        probe = self._approve_only({"bob"})
+        state = sop.compute_ack_state(
+            comments, "alice", self.items, self.aliases, probe, high_risk=False
+        )
+        self.assertEqual(state["no-backwards-compat"]["ackers"], ["bob"])
+
+    def test_engineers_ack_alone_fails_root_cause_when_high_risk(self):
+        # High-risk PR: only ceo can ack. Engineers-only ack must fail.
+        comments = [_comment("bob", "/sop-ack root-cause")]
+        # Probe: bob is in engineers, not ceo. Under high_risk,
+        # required_teams_high_risk=[ceo] → bob is NOT approved.
+        # Probe receives the items + flag indirectly via main(); for
+        # the unit-test path we inject a probe that rejects bob.
+        probe = self._approve_only(set())  # nobody is in ceo
+        state = sop.compute_ack_state(
+            comments, "alice", self.items, self.aliases, probe, high_risk=True
+        )
+        self.assertEqual(state["root-cause"]["ackers"], [])
+        self.assertIn("bob", state["root-cause"]["rejected"]["not_in_team"])
+
+    def test_ceo_ack_satisfies_root_cause_when_high_risk(self):
+        # High-risk PR + ceo-team approver → passes (the senior path).
+        comments = [_comment("hongming", "/sop-ack root-cause")]
+        probe = self._approve_only({"hongming"})
+        state = sop.compute_ack_state(
+            comments, "alice", self.items, self.aliases, probe, high_risk=True
+        )
+        self.assertEqual(state["root-cause"]["ackers"], ["hongming"])
+
+    def test_self_ack_still_forbidden_even_with_widened_eligibility(self):
+        # Author cannot self-ack — widening teams must NOT weaken
+        # the non-author rule.
+        comments = [_comment("alice", "/sop-ack root-cause")]
+        probe = self._approve_only({"alice"})
+        state = sop.compute_ack_state(
+            comments, "alice", self.items, self.aliases, probe, high_risk=False
+        )
+        self.assertEqual(state["root-cause"]["ackers"], [])
+        self.assertIn("alice", state["root-cause"]["rejected"]["self_ack"])
+
+
+class TestHighRiskClassUsesElevatedListInConfig(unittest.TestCase):
+    """End-to-end: the shipped config + RFC#450 predicate must keep
+    root-cause / no-backwards-compat gated on ceo for high-risk PRs."""
+
+    def test_root_cause_high_risk_elevated_to_ceo_only(self):
+        items = _items_by_slug()
+        # tier:high alone makes the PR high-risk → root-cause needs ceo.
+        self.assertEqual(
+            sop.resolve_required_teams(items["root-cause"], high_risk=True),
+            ["ceo"],
+        )
+        # Default class accepts engineers/managers/ceo.
+        self.assertEqual(
+            sorted(sop.resolve_required_teams(items["root-cause"], high_risk=False)),
+            sorted(["engineers", "managers", "ceo"]),
+        )
+
+    def test_no_backwards_compat_high_risk_elevated_to_ceo_only(self):
+        items = _items_by_slug()
+        self.assertEqual(
+            sop.resolve_required_teams(items["no-backwards-compat"], high_risk=True),
+            ["ceo"],
+        )
+        self.assertEqual(
+            sorted(sop.resolve_required_teams(items["no-backwards-compat"], high_risk=False)),
+            sorted(["engineers", "managers", "ceo"]),
+        )
+
+    def test_other_items_unchanged_by_risk_class(self):
+        # Items without required_teams_high_risk are unaffected.
+        items = _items_by_slug()
+        for slug in (
+            "comprehensive-testing",
+            "local-postgres-e2e",
+            "staging-smoke",
+            "five-axis-review",
+            "memory-consulted",
+        ):
+            self.assertEqual(
+                sop.resolve_required_teams(items[slug], high_risk=False),
+                sop.resolve_required_teams(items[slug], high_risk=True),
+                f"item {slug} should not be affected by risk-class",
+            )
+
+
+# ---------------------------------------------------------------------------
+# get_issue_comments — streaming + minimal-dict shape (task #369 / OOM fix)
+# ---------------------------------------------------------------------------
+
+
+class _FakeReq:
+    """Stand-in for GiteaClient._req that serves canned pages."""
+
+    def __init__(self, pages):
+        # pages: list[list[dict]]; one page per call, exhausted in order.
+        self._pages = list(pages)
+        self.calls = []
+
+    def __call__(self, method, path, body=None, ok_codes=(200, 201, 204)):
+        self.calls.append((method, path))
+        if not self._pages:
+            return 200, []
+        return 200, self._pages.pop(0)
+
+
+class TestGetIssueCommentsStreaming(unittest.TestCase):
+    """Verify the OOM-fix invariants — minimal-dict shape + page break."""
+
+    def _client_with_pages(self, pages):
+        client = sop.GiteaClient("git.example.com", "tok")
+        client._req = _FakeReq(pages)  # type: ignore[method-assign]
+        return client
+
+    def test_minimal_dict_shape_drops_large_fields(self):
+        """get_issue_comments must DROP html_url/assets/timestamps/etc. and
+        keep ONLY {user.login, body} — that's the whole OOM-prevention."""
+        full_page = [
+            {
+                "id": 1234,
+                "html_url": "https://example.com/some-huge-url",
+                "pull_request_url": "https://example.com/some-other-huge-url",
+                "issue_url": "https://example.com/yet-another-url",
+                "user": {"login": "bob", "avatar_url": "x" * 4000, "id": 99},
+                "original_author": "",
+                "original_author_id": 0,
+                "body": "/sop-ack comprehensive-testing\n\nlooks good",
+                "assets": ["x" * 1000, "y" * 1000],
+                "created_at": "2026-05-19T01:02:03Z",
+                "updated_at": "2026-05-19T01:02:03Z",
+            }
+        ]
+        client = self._client_with_pages([full_page])
+        out = client.get_issue_comments("o", "r", 1)
+        self.assertEqual(len(out), 1)
+        # Only the two whitelisted keys + nested user.login
+        self.assertEqual(set(out[0].keys()), {"user", "body"})
+        self.assertEqual(set(out[0]["user"].keys()), {"login"})
+        self.assertEqual(out[0]["user"]["login"], "bob")
+        self.assertEqual(out[0]["body"], "/sop-ack comprehensive-testing\n\nlooks good")
+        # Critical: avatar/assets/timestamps/etc. must be gone (~4KB+ each).
+        self.assertNotIn("html_url", out[0])
+        self.assertNotIn("assets", out[0])
+        self.assertNotIn("created_at", out[0])
+
+    def test_pagination_break_on_short_page(self):
+        # Page-size 50; a page of <50 means no more pages.
+        page1 = [{"user": {"login": "u"}, "body": "x"}] * 7
+        client = self._client_with_pages([page1])
+        out = client.get_issue_comments("o", "r", 2)
+        self.assertEqual(len(out), 7)
+        # Should have made exactly 1 _req call (no page-2 probe).
+        self.assertEqual(len(client._req.calls), 1)
+
+    def test_pagination_continues_until_empty(self):
+        # Two full pages + one short page.
+        page1 = [{"user": {"login": "u"}, "body": "x"}] * 50
+        page2 = [{"user": {"login": "u"}, "body": "y"}] * 50
+        page3 = [{"user": {"login": "u"}, "body": "z"}] * 3
+        client = self._client_with_pages([page1, page2, page3])
+        out = client.get_issue_comments("o", "r", 3)
+        self.assertEqual(len(out), 103)
+        self.assertEqual(len(client._req.calls), 3)
+
+    def test_max_comments_caps_collection(self):
+        page1 = [{"user": {"login": "u"}, "body": "x"}] * 50
+        page2 = [{"user": {"login": "u"}, "body": "y"}] * 50
+        page3 = [{"user": {"login": "u"}, "body": "z"}] * 50
+        client = self._client_with_pages([page1, page2, page3])
+        out = client.get_issue_comments("o", "r", 4, max_comments=75)
+        self.assertEqual(len(out), 75)
+        # Stops short: shouldn't have requested page-3.
+        self.assertLessEqual(len(client._req.calls), 2)
+
+    def test_oversized_body_truncated(self):
+        # An individual comment with a multi-MiB body (e.g. pasted CI log)
+        # must NOT pull the whole thing into memory. The directive parser
+        # only needs the first ~8 KiB to find /sop-* markers.
+        huge_body = "/sop-ack comprehensive-testing\n" + ("X" * (4 * 1024 * 1024))
+        page = [{"user": {"login": "bob"}, "body": huge_body}]
+        client = self._client_with_pages([page])
+        out = client.get_issue_comments("o", "r", 99)
+        self.assertEqual(len(out), 1)
+        # Cap is 8 KiB; comment body must be <= 8 KiB after streaming.
+        self.assertLessEqual(len(out[0]["body"]), 8 * 1024)
+        # Marker still discoverable at the start.
+        self.assertTrue(out[0]["body"].startswith("/sop-ack comprehensive-testing"))
+
+    def test_iter_handles_missing_user_or_body(self):
+        # Defensive: Gitea has been seen to return user=null on deleted users.
+        page = [
+            {"user": None, "body": "abandoned-author"},
+            {"user": {"login": "alice"}, "body": None},
+            {"body": "no-user-key"},
+            {"user": {"login": "bob"}, "body": "ok"},
+        ]
+        client = self._client_with_pages([page])
+        out = client.get_issue_comments("o", "r", 5)
+        self.assertEqual(len(out), 4)
+        self.assertEqual(out[0]["user"]["login"], "")
+        self.assertEqual(out[0]["body"], "abandoned-author")
+        self.assertEqual(out[1]["user"]["login"], "alice")
+        self.assertEqual(out[1]["body"], "")
+        self.assertEqual(out[2]["user"]["login"], "")
+        self.assertEqual(out[3]["user"]["login"], "bob")
+
+    def test_minimal_dicts_work_with_compute_ack_state(self):
+        """Round-trip: minimal dicts feed back through compute_ack_state."""
+        page = [{"user": {"login": "bob"}, "body": "/sop-ack comprehensive-testing"}]
+        client = self._client_with_pages([page])
+        comments = client.get_issue_comments("o", "r", 6)
+        items = _items_by_slug()
+        aliases = _numeric_aliases()
+        state = sop.compute_ack_state(
+            comments, "alice", items, aliases, lambda slug, users: list(users)
+        )
+        self.assertEqual(state["comprehensive-testing"]["ackers"], ["bob"])
+
+
+# ---------------------------------------------------------------------------
+# probe() na-gate fallback — fix for #355-class KeyError 'security-review'
+# ---------------------------------------------------------------------------
+
+
+class TestComputeNaStateAcceptsGateNotInItems(unittest.TestCase):
+    """compute_na_state passes the gate NAME to probe(); when the gate is
+    NOT also an items entry (the common case for `security-review`,
+    `qa-review`), probe must fall back to the gate's own required_teams
+    instead of KeyError'ing on items_by_slug[slug].
+
+    This test exercises the public surface (compute_na_state) rather than
+    the inline `probe` closure, because the closure is built inside main().
+    We simulate the fallback by passing a probe that mirrors the production
+    contract — slug may be either an item OR an n/a-gate key, both are valid.
+    """
+
+    def test_na_gate_with_required_teams_resolves_without_keyerror(self):
+        na_gates = {
+            "security-review": {
+                "required_teams": ["security", "managers", "ceo"],
+                "description": "security N/A",
+            },
+        }
+        comments = [
+            {"user": {"login": "carol"}, "body": "/sop-n/a security-review docs-only"},
+        ]
+        # Probe approves any user in the security team; importantly it does
+        # NOT try items_by_slug[slug] for the gate name.
+        called_with = []
+
+        def probe(slug, users):
+            called_with.append(slug)
+            # production probe accepts gate-name OR item-slug; for this test
+            # we just approve everyone.
+            return list(users)
+
+        na_state = sop.compute_na_state(comments, "alice", na_gates, probe)
+        self.assertTrue(na_state["security-review"]["declared"])
+        self.assertEqual(na_state["security-review"]["decl_ackers"], ["carol"])
+        # probe must have been called with the GATE name, not an item slug.
+        self.assertEqual(called_with, ["security-review"])
+
+    def test_na_gate_self_declaration_rejected(self):
+        # Author cannot self-declare N/A — pre-existing invariant; pin it
+        # so the new probe-fallback doesn't regress this.
+        na_gates = {"security-review": {"required_teams": ["security"]}}
+        comments = [
+            {"user": {"login": "alice"}, "body": "/sop-n/a security-review"},
+        ]
+        na_state = sop.compute_na_state(
+            comments, "alice", na_gates, lambda *_: ["alice"]
+        )
+        self.assertFalse(na_state["security-review"]["declared"])
@@ -50,6 +50,34 @@ tier_failure_mode:
  "tier:low": soft
 default_mode: hard  # used when no tier:* label is present

+# High-risk class (RFC#450 Option C, governance-fix for internal#442).
+#
+# A PR is "high-risk" when ANY of the listed labels are applied OR when
+# the PR has `tier:high` (mechanically the strictest existing tier).
+# High-risk items use `required_teams_high_risk` (when present on the
+# item); non-high-risk items use the default `required_teams`.
+#
+# This closes the inconsistency that the SOP charter already mandates
+# `tier:high → ceo only` for the sibling `sop-tier-check` gate; the
+# sop-checklist's `root-cause` and `no-backwards-compat` items now
+# follow the same risk-classed two-eyes shape:
+#   - Default class (tier:low/medium, not high-risk): a non-author
+#     engineers/managers/ceo ack satisfies the item — 25+ live
+#     identities, no dependency on a dead/inactive senior persona
+#     token.
+#   - High-risk class (tier:high OR any high_risk_label): still
+#     requires a non-author ceo ack (durable human team).
+#
+# Tightening: add labels to high_risk_labels.
+# Loosening: remove labels.
+high_risk_labels:
+  - "risk:high"
+  - "area:security"
+  - "area:schema"
+  - "area:fleet-image"
+  - "area:identity"
+  - "area:gate-meta"
+
 items:
  - slug: comprehensive-testing
    numeric_alias: 1
@@ -78,11 +106,15 @@ items:
  - slug: root-cause
    numeric_alias: 4
    pr_section_marker: "Root-cause not symptom"
-    required_teams: [managers, ceo]
+    required_teams: [engineers, managers, ceo]
+    required_teams_high_risk: [ceo]
    description: >-
-      One-sentence root-cause statement. Ack from managers tier
-      (team-leads) or ceo. Senior judgment required to attest
-      root-cause-versus-symptom.
+      One-sentence root-cause statement. Default class: non-author
+      engineers/managers/ceo ack suffices (engineers can attest
+      root-cause-vs-symptom for routine fixes). High-risk class
+      (see `high_risk_labels`): non-author ceo ack required —
+      senior judgment for irreversible/security/identity/gate
+      changes. Closes internal#442 + tracks RFC#450.

  - slug: five-axis-review
    numeric_alias: 5
@@ -95,10 +127,14 @@ items:
  - slug: no-backwards-compat
    numeric_alias: 6
    pr_section_marker: "No backwards-compat shim / dead code added"
-    required_teams: [managers, ceo]
+    required_teams: [engineers, managers, ceo]
+    required_teams_high_risk: [ceo]
    description: >-
-      Yes/no + justification if no. Senior ack required because
-      backward-compat shims are how dead-code accretes.
+      Yes/no + justification if no. Default class: non-author
+      engineers/managers/ceo ack suffices. High-risk class
+      (see `high_risk_labels`): non-author ceo ack required —
+      senior judgment for shim-versus-real-fix on irreversible
+      surfaces. Closes internal#442 + tracks RFC#450.

  - slug: memory-consulted
    numeric_alias: 7
@@ -158,8 +158,68 @@ jobs:
            echo "NOTE: No warning in output (may be suppressed by log level)"
          fi

+      - name: Reproduce openclaw failure — pipe held OPEN, no EOF
+        run: |
+          set -euo pipefail
+          echo "=== keep-stdin-open pipe (the real openclaw / Claude Code case) ==="
+          echo ""
+          echo "Before the readline() fix this HANGS: main() did"
+          echo "  stdin.read(65536)  -> on a pipe, blocks until 64KB OR EOF."
+          echo "An MCP client sends one ~150B initialize and keeps stdin"
+          echo "open waiting for the response, so the server never parsed"
+          echo "the request and the client timed out (openclaw: 'MCP error"
+          echo "-32000: Connection closed'). The earlier regular-file /"
+          echo "heredoc-pipe steps PASSED through this bug because a file"
+          echo "(or a closing heredoc) yields EOF immediately."
+          echo ""
+
+          # Drive the server through a real pipe that stays OPEN: write
+          # one initialize, do NOT close stdin, and require a response
+          # within a hard timeout. read(65536) -> no output -> timeout
+          # kills it -> FAIL. readline() -> immediate response -> PASS.
+          python - <<'PYEOF'
+          import json, subprocess, sys, time, select
+
+          proc = subprocess.Popen(
+              [sys.executable, "a2a_mcp_server.py"],
+              stdin=subprocess.PIPE, stdout=subprocess.PIPE,
+              stderr=subprocess.STDOUT,
+              env={**__import__("os").environ},
+          )
+          req = json.dumps({
+              "jsonrpc": "2.0", "id": 1, "method": "initialize",
+              "params": {"protocolVersion": "2024-11-05",
+                         "capabilities": {},
+                         "clientInfo": {"name": "keepopen", "version": "1"}},
+          }) + "\n"
+          proc.stdin.write(req.encode())
+          proc.stdin.flush()
+          # Deliberately DO NOT close proc.stdin — mirror a live MCP client.
+
+          deadline = time.time() + 15
+          line = b""
+          while time.time() < deadline:
+              r, _, _ = select.select([proc.stdout], [], [], 1)
+              if r:
+                  line = proc.stdout.readline()
+                  if line:
+                      break
+          proc.kill()
+
+          if not line:
+              print("FAIL: no response within 15s on an open pipe — "
+                    "stdin.read(65536) regression is back")
+              sys.exit(1)
+          resp = json.loads(line.decode())
+          assert resp.get("id") == 1 and "result" in resp, \
+              f"unexpected response: {line[:200]!r}"
+          assert resp["result"]["serverInfo"]["name"] == "molecule", \
+              f"wrong serverInfo: {line[:200]!r}"
+          print("PASS: server answered initialize on a still-open pipe")
+          PYEOF
+
      - name: Run unit tests for stdio transport
        run: |
          set -euo pipefail
          echo "=== Running stdio transport unit tests ==="
-          python -m pytest tests/test_a2a_mcp_server.py::TestStdioPipeAssertion -v --no-cov
+          python -m pytest tests/test_a2a_mcp_server.py::TestStdioPipeAssertion tests/test_a2a_mcp_server.py::TestStdioKeepOpenPipe -v --no-cov
@@ -319,7 +319,7 @@ jobs:
        with:
          node-version: '22'
      - if: always()
-        run: rm -f package-lock.json && npm install
+        run: npm ci --include=optional --prefer-offline
      - if: always()
        run: npm run build
      - if: always()
@@ -401,7 +401,7 @@ jobs:

  canvas-deploy-reminder:
    name: Canvas Deploy Reminder
-    runs-on: ubuntu-latest
+    runs-on: docker-host
    # mc#774 root-fix: added job-level `if:` so ci-required-drift.py's
    # ci_job_names() detects this as github.ref-gated and skips it from F1.
    # The step-level exit 0 handles the "not main push" case; the job-level
@@ -538,11 +538,13 @@ jobs:
  all-required:
    # Aggregator sentinel — RFC internal#219 §2 (Phase 4 — closes internal#286).
    #
-    # Single stable required-status name that branch protection points at;
-    # CI churns underneath in `needs:` without any protection edits. Mirrors
-    # the molecule-controlplane Phase 2a impl shipped in CP PR#112 and
-    # referenced by `internal#286` ("Phase 4 is a single small PR... mirrors
-    # CP's existing one").
+    # Emits `CI / all-required (<event>)` where <event> is the workflow trigger
+    # (e.g. `CI / all-required (pull_request)`, `CI / all-required (push)`).
+    # Branch protection MUST be updated to require the event-suffixed name —
+    # requiring `CI / all-required` (bare, no suffix) silently blocks all merges
+    # because Gitea treats absent status contexts as pending (not skipped), and
+    # no workflow emits the bare name. Fixed: BP now requires
+    # `CI / all-required (pull_request)` per issue #1473.
    #
    # Closes the failure mode where status_check_contexts on molecule-core/main
    # only listed `Secret scan` + `sop-tier-check` (the 2 meta-gates), so real
@@ -43,6 +43,18 @@ name: Continuous synthetic E2E (staging)

 on:
  schedule:
+    # Every 30 minutes, on :02 and :32. This keeps a recurring SaaS
+    # behavior probe while cutting runner occupancy from this workflow by
+    # roughly two thirds; fast liveness belongs in the lighter smoke/heartbeat
+    # probes, not in a full tenant/workspace synth every 10 minutes.
+    #
+    # Previous cadence was every 10 minutes (:02 :12 :22 :32 :42 :52).
+    # The current operator-host runner pool is the bottleneck, so full
+    # synth E2E is deliberately lower-cadence until it moves to a dedicated
+    # runner host or warm-runtime pool.
+    #
+    # Historical notes from the 10-minute shape:
+    #
    # Every 10 minutes, on :02 :12 :22 :32 :42 :52. Three constraints:
    #   1. Stay off the top-of-hour. GitHub Actions scheduler drops
    #      :00 firings under high load (own docs:
@@ -66,7 +78,7 @@ on:
    # fires = ~30 min cadence; closer to the 20-min target than the
    # current shape and provides a real degradation alarm if drops
    # get worse.
-    - cron: '2,12,22,32,42,52 * * * *'
+    - cron: '2,32 * * * *'
 permissions:
  contents: read
  # No issue-write here — failures surface as red runs in the workflow
@@ -108,7 +108,20 @@ env:

 jobs:
  detect-changes:
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: pin to `docker-host` so the e2e-api lane lands
+    # on Linux operator-host runners (molecule-runner-*) that carry the
+    # `molecule-core-net` bridge network + a working `aws ecr get-login-
+    # password | docker login` path. The bare `ubuntu-latest` label is
+    # also accepted by hongming-pc-runner-* (Windows act_runner v1.0.3),
+    # where the docker.sock-bound steps below fail non-deterministically
+    # (e.g. `docker run -d --name pg-e2e-api-...` with port-bind +
+    # `docker exec ... pg_isready` cannot work against a Windows daemon).
+    # detect-changes itself doesn't bind docker.sock, but pinning here too
+    # keeps both jobs on the same lane so we don't re-roll the dice on
+    # workspace-volume cross-host surprises and the routing rule is
+    # discoverable in one place. Mirror of mc#1543 (handlers-postgres-
+    # integration). See internal#512 for the class defect.
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -160,7 +173,10 @@ jobs:
  e2e-api:
    needs: detect-changes
    name: E2E API Smoke Test
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: must run on operator-host Linux runners (where
+    # docker.sock + `molecule-core-net` + `aws ecr ...` work). See
+    # detect-changes for the full rationale.
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -365,6 +381,9 @@ jobs:
      - name: Run poll-mode chat upload E2E (RFC #2891)
        if: needs.detect-changes.outputs.api == 'true'
        run: bash tests/e2e/test_poll_mode_chat_upload_e2e.sh
+      - name: Run today's-PR-coverage E2E (mc#1525/1535/1536/1539/1542 fix-specific assertions)
+        if: needs.detect-changes.outputs.api == 'true'
+        run: bash tests/e2e/test_today_pr_coverage_e2e.sh
      - name: Dump platform log on failure
        if: failure() && needs.detect-changes.outputs.api == 'true'
        run: cat workspace-server/platform.log || true
@@ -1,8 +1,10 @@
 name: E2E Chat

 # Comprehensive Playwright E2E for the unified chat stack (desktop
-# ChatTab + mobile MobileChat). Runs on every PR that touches canvas,
-# workspace-server, or this workflow file.
+# ChatTab + mobile MobileChat). Heavy browser execution is intentionally
+# outside the normal required PR path: PRs run it only after entering the
+# `merge-queue`, while push/main, nightly, and manual dispatch preserve
+# coverage without making every PR pay the full runtime/browser cost.
 #
 # Architecture:
 #   1. Ephemeral Postgres + Redis (docker, unique container names)
@@ -22,6 +24,11 @@ on:
    branches: [main, staging]
  pull_request:
    branches: [main, staging]
+  schedule:
+    # Nightly at 09:00 UTC. Keeps coverage for the currently non-required
+    # heavy browser lane without spending runner time on every PR.
+    - cron: '0 9 * * *'
+  workflow_dispatch:

 concurrency:
  group: e2e-chat-${{ github.event.pull_request.head.sha || github.sha }}
@@ -33,7 +40,13 @@ env:
 jobs:
  # bp-exempt: helper job; real gate is E2E Chat / E2E Chat (pull_request)
  detect-changes:
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: pin to `docker-host` (Linux operator-host
+    # runners). The bare `ubuntu-latest` label is also advertised by
+    # hongming-pc-runner-* (Windows act_runner v1.0.3) where the
+    # docker.sock-bound steps below fail. Mirror of mc#1543
+    # (handlers-postgres-integration). See internal#512 for the class
+    # defect.
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -44,7 +57,14 @@ jobs:
        with:
          fetch-depth: 0
      - id: decide
+        env:
+          GITEA_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          QUEUE_LABEL: merge-queue
        run: |
+          if [ "${{ github.event_name }}" = "schedule" ] || [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            echo "chat=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
          BASE="${GITHUB_BASE_REF:-${{ github.event.before }}}"
          if [ "${{ github.event_name }}" = "pull_request" ] && [ -n "${{ github.event.pull_request.base.sha }}" ]; then
            BASE="${{ github.event.pull_request.base.sha }}"
@@ -61,9 +81,26 @@ jobs:
            exit 0
          fi
          CHANGED=$(git diff --name-only "$BASE" HEAD)
-          if echo "$CHANGED" | grep -qE '^(canvas/|workspace-server/|\.gitea/workflows/e2e-chat\.yml$)'; then
+          if ! echo "$CHANGED" | grep -qE '^(canvas/|workspace-server/|\.gitea/workflows/e2e-chat\.yml$)'; then
+            echo "chat=false" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          if [ "${{ github.event_name }}" != "pull_request" ]; then
+            echo "chat=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+
+          authfile=$(mktemp)
+          chmod 600 "$authfile"
+          printf 'header = "Authorization: token %s"\n' "$GITEA_TOKEN" > "$authfile"
+          labels=$(curl -fsS -K "$authfile" \
+            "${{ github.server_url }}/api/v1/repos/${{ github.repository }}/issues/${{ github.event.pull_request.number }}/labels" \
+            | python3 -c 'import json,sys; print("\n".join(label.get("name","") for label in json.load(sys.stdin)))')
+          rm -f "$authfile"
+          if printf '%s\n' "$labels" | grep -qx "$QUEUE_LABEL"; then
            echo "chat=true" >> "$GITHUB_OUTPUT"
          else
+            echo "PR is not in merge-queue; skipping heavy E2E Chat for normal PR path."
            echo "chat=false" >> "$GITHUB_OUTPUT"
          fi

@@ -71,7 +108,9 @@ jobs:
  e2e-chat:
    needs: detect-changes
    name: E2E Chat
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: docker run/exec for postgres + redis containers.
+    # Must land on operator-host Linux (docker-host).
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -222,7 +261,14 @@ jobs:
      - name: Install Playwright browsers
        if: needs.detect-changes.outputs.chat == 'true'
        working-directory: canvas
-        run: npx playwright install --with-deps chromium
+        run: |
+          PREBAKED_PLAYWRIGHT=/ms-playwright
+          if [ -d "${PREBAKED_PLAYWRIGHT}" ] && find "${PREBAKED_PLAYWRIGHT}" -maxdepth 3 -type f -name 'chrome' | grep -q .; then
+            echo "Using prebaked Playwright Chromium from ${PREBAKED_PLAYWRIGHT}"
+            echo "PLAYWRIGHT_BROWSERS_PATH=${PREBAKED_PLAYWRIGHT}" >> "$GITHUB_ENV"
+            exit 0
+          fi
+          npx playwright install --with-deps chromium

      - name: Start canvas dev server (background)
        if: needs.detect-changes.outputs.chat == 'true'
@@ -52,6 +52,30 @@ name: E2E Peer Visibility (literal MCP list_peers)
 #     flip-to-required-ready (mirrors e2e-staging-saas.yml's proven shape;
 #     real EC2-provisioning E2E is push/dispatch/cron only — it is 30+ min
 #     and cannot run per-PR-update).
+#
+# LOCAL BACKEND (added 2026-05-15 — feedback_local_must_mimic_production,
+# feedback_mandatory_local_e2e_before_ship, feedback_local_test_before_
+# staging_e2e)
+# --------------------------------------------------------------------
+# The standing rule is that the local prod-mimic stack runs a MANDATORY
+# local-Postgres E2E BEFORE staging E2E. A staging-only peer-visibility
+# gate caught regressions late + expensively (cold EC2). The
+# `peer-visibility-local` job below runs the SAME byte-identical
+# assertion (tests/e2e/lib/peer_visibility_assert.sh) against the local
+# docker-compose stack — built + booted exactly like e2e-api.yml's
+# proven E2E API Smoke Test job (ephemeral pg/redis ports, go build,
+# background platform-server). It runs on PR + push (local boot is
+# minutes, not the 30+ min cold-EC2 path), so peer-visibility is part of
+# the local gate that fires before the staging E2E.
+#
+# It is its OWN non-required status context `E2E Peer Visibility (local)`
+# — same non-required-by-design decision as the staging job (red until
+# Hermes-401 #162 / OpenClaw-never-online #165 land; flip-to-required
+# tracked at molecule-core#1296). It is an HONEST gate: NO
+# continue-on-error mask (feedback_fix_root_not_symptom). It is kept a
+# distinct context (not folded into e2e-api.yml's required `E2E API
+# Smoke Test`) precisely so a deliberately-RED-today gate cannot wedge
+# the required local-E2E job or any unrelated merge.

 on:
  push:
@@ -65,6 +89,8 @@ on:
      - 'workspace/a2a_mcp_server.py'
      - 'workspace/platform_tools/registry.py'
      - 'tests/e2e/test_peer_visibility_mcp_staging.sh'
+      - 'tests/e2e/test_peer_visibility_mcp_local.sh'
+      - 'tests/e2e/lib/peer_visibility_assert.sh'
      - '.gitea/workflows/e2e-peer-visibility.yml'
  pull_request:
    branches: [main]
@@ -77,6 +103,8 @@ on:
      - 'workspace/a2a_mcp_server.py'
      - 'workspace/platform_tools/registry.py'
      - 'tests/e2e/test_peer_visibility_mcp_staging.sh'
+      - 'tests/e2e/test_peer_visibility_mcp_local.sh'
+      - 'tests/e2e/lib/peer_visibility_assert.sh'
      - '.gitea/workflows/e2e-peer-visibility.yml'
  workflow_dispatch:
  schedule:
@@ -108,16 +136,160 @@ jobs:
    timeout-minutes: 5
    steps:
      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - name: Validate driving script
+      - name: Validate driving scripts + shared assertion lib
        run: |
+          bash -n tests/e2e/lib/peer_visibility_assert.sh
+          echo "lib/peer_visibility_assert.sh — bash syntax OK"
          bash -n tests/e2e/test_peer_visibility_mcp_staging.sh
          echo "test_peer_visibility_mcp_staging.sh — bash syntax OK"
-          echo "Real fresh-provision MCP list_peers E2E runs on push to"
+          bash -n tests/e2e/test_peer_visibility_mcp_local.sh
+          echo "test_peer_visibility_mcp_local.sh — bash syntax OK"
+          echo "Staging fresh-provision MCP list_peers E2E runs on push to"
          echo "main / workflow_dispatch / daily cron (30+ min EC2 boot)."
+          echo "The LOCAL backend runs in the peer-visibility-local job"
+          echo "below on this same PR (local docker-compose stack)."

-  # Real gate: provisions a throwaway org + sibling-per-runtime, drives
-  # the LITERAL list_peers MCP call per runtime, asserts 200 + expected
-  # peer set, then scoped teardown. push(main)/dispatch/cron only.
+  # LOCAL gate: same byte-identical assertion against the local prod-mimic
+  # docker-compose stack — the MANDATORY local-E2E that must run BEFORE
+  # the staging E2E (feedback_mandatory_local_e2e_before_ship,
+  # feedback_local_test_before_staging_e2e). Bootstrap mirrors
+  # e2e-api.yml's proven E2E API Smoke Test job (per-run container names +
+  # ephemeral host ports so concurrent host-network act_runner runs don't
+  # collide; go build; background platform-server). Its OWN non-required
+  # status context `E2E Peer Visibility (local)` — non-required-by-design
+  # exactly like the staging job (red until #162/#165 land;
+  # flip-to-required tracked at molecule-core#1296). HONEST gate, NO
+  # continue-on-error mask (feedback_fix_root_not_symptom). Runs on PR +
+  # push (local boot is minutes, not the 30+ min cold-EC2 path).
+  # bp-required: pending #1296
+  peer-visibility-local:
+    name: E2E Peer Visibility (local)
+    runs-on: docker-host
+    timeout-minutes: 30
+    env:
+      # Per-run names + ephemeral ports — same collision-avoidance as
+      # e2e-api.yml (host-network act_runner; feedback_act_runner_*).
+      PG_CONTAINER: pg-e2e-pv-${{ github.run_id }}-${{ github.run_attempt }}
+      REDIS_CONTAINER: redis-e2e-pv-${{ github.run_id }}-${{ github.run_attempt }}
+      # LLM keys so hermes/openclaw can actually boot. The local script
+      # SKIPs (not fails) any runtime whose key is absent, so a partially
+      # keyed CI env still exercises whatever it can.
+      CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.E2E_CLAUDE_CODE_OAUTH_TOKEN }}
+      E2E_MINIMAX_API_KEY: ${{ secrets.MOLECULE_STAGING_MINIMAX_API_KEY }}
+      E2E_ANTHROPIC_API_KEY: ${{ secrets.MOLECULE_STAGING_ANTHROPIC_API_KEY }}
+      E2E_OPENAI_API_KEY: ${{ secrets.MOLECULE_STAGING_OPENAI_API_KEY }}
+      PV_RUNTIMES: "hermes openclaw claude-code"
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
+        with:
+          go-version: 'stable'
+          cache: true
+          cache-dependency-path: workspace-server/go.sum
+      - name: Pre-pull alpine + ensure provisioner network
+        run: |
+          docker pull alpine:latest >/dev/null
+          docker network create molecule-core-net >/dev/null 2>&1 || true
+          echo "alpine:latest pre-pulled; molecule-core-net ensured."
+      - name: Start Postgres (docker, ephemeral port)
+        run: |
+          docker rm -f "$PG_CONTAINER" 2>/dev/null || true
+          docker run -d --name "$PG_CONTAINER" \
+            -e POSTGRES_USER=dev -e POSTGRES_PASSWORD=dev -e POSTGRES_DB=molecule \
+            -p 0:5432 postgres:16 >/dev/null
+          PG_PORT=$(docker port "$PG_CONTAINER" 5432/tcp | awk -F: '/^0\.0\.0\.0:/ {print $2; exit}')
+          [ -n "$PG_PORT" ] || PG_PORT=$(docker port "$PG_CONTAINER" 5432/tcp | head -1 | awk -F: '{print $NF}')
+          if [ -z "$PG_PORT" ]; then
+            echo "::error::Could not resolve host port for $PG_CONTAINER"
+            docker logs "$PG_CONTAINER" || true; exit 1
+          fi
+          echo "DATABASE_URL=postgres://dev:dev@127.0.0.1:${PG_PORT}/molecule?sslmode=disable" >> "$GITHUB_ENV"
+          for i in $(seq 1 30); do
+            docker exec "$PG_CONTAINER" pg_isready -U dev >/dev/null 2>&1 && { echo "Postgres ready after ${i}s"; exit 0; }
+            sleep 1
+          done
+          echo "::error::Postgres did not become ready in 30s"; docker logs "$PG_CONTAINER" || true; exit 1
+      - name: Start Redis (docker, ephemeral port)
+        run: |
+          docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true
+          docker run -d --name "$REDIS_CONTAINER" -p 0:6379 redis:7 >/dev/null
+          REDIS_PORT=$(docker port "$REDIS_CONTAINER" 6379/tcp | awk -F: '/^0\.0\.0\.0:/ {print $2; exit}')
+          [ -n "$REDIS_PORT" ] || REDIS_PORT=$(docker port "$REDIS_CONTAINER" 6379/tcp | head -1 | awk -F: '{print $NF}')
+          if [ -z "$REDIS_PORT" ]; then
+            echo "::error::Could not resolve host port for $REDIS_CONTAINER"
+            docker logs "$REDIS_CONTAINER" || true; exit 1
+          fi
+          echo "REDIS_URL=redis://127.0.0.1:${REDIS_PORT}" >> "$GITHUB_ENV"
+          for i in $(seq 1 15); do
+            docker exec "$REDIS_CONTAINER" redis-cli ping 2>/dev/null | grep -q PONG && { echo "Redis ready after ${i}s"; exit 0; }
+            sleep 1
+          done
+          echo "::error::Redis did not become ready in 15s"; docker logs "$REDIS_CONTAINER" || true; exit 1
+      - name: Build platform
+        working-directory: workspace-server
+        run: go build -o platform-server ./cmd/server
+      - name: Pick platform port
+        run: |
+          PLATFORM_PORT=$(python3 - <<'PY'
+          import socket
+          with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+              s.bind(("127.0.0.1", 0))
+              print(s.getsockname()[1])
+          PY
+          )
+          echo "PORT=${PLATFORM_PORT}" >> "$GITHUB_ENV"
+          echo "BASE=http://127.0.0.1:${PLATFORM_PORT}" >> "$GITHUB_ENV"
+          echo "Platform host port: ${PLATFORM_PORT}"
+      - name: Kill stale platform-server before start
+        run: |
+          killed=0
+          for pid in $(grep -l "platform-serve" /proc/[0-9]*/comm 2>/dev/null); do
+            kpid="${pid%/comm}"; kpid="${kpid##*/}"
+            cmdline=$(cat "/proc/${kpid}/cmdline" 2>/dev/null | tr '\0' ' ')
+            if echo "$cmdline" | grep -q "platform-server"; then
+              echo "Killing stale platform-server pid ${kpid}"
+              kill "$kpid" 2>/dev/null || true; killed=$((killed + 1))
+            fi
+          done
+          [ "$killed" -gt 0 ] && sleep 2 || true
+          echo "stale-kill done ($killed killed)"
+      - name: Start platform (background)
+        working-directory: workspace-server
+        run: |
+          ./platform-server > platform.log 2>&1 &
+          echo $! > platform.pid
+      - name: Wait for /health
+        run: |
+          for i in $(seq 1 30); do
+            curl -sf "$BASE/health" > /dev/null && { echo "Platform up after ${i}s"; exit 0; }
+            sleep 1
+          done
+          echo "::error::Platform did not become healthy in 30s"
+          cat workspace-server/platform.log || true; exit 1
+      - name: Run LOCAL fresh-provision peer-visibility E2E (literal MCP list_peers)
+        # HONEST gate — NO continue-on-error. Red today (Hermes-401 #162 /
+        # OpenClaw-never-online #165 not yet fixed); green when they land.
+        # Non-required-by-design via its distinct status context until the
+        # molecule-core#1296 flip-to-required.
+        run: bash tests/e2e/test_peer_visibility_mcp_local.sh
+      - name: Dump platform log on failure
+        if: failure()
+        run: cat workspace-server/platform.log || true
+      - name: Stop platform
+        if: always()
+        run: |
+          if [ -f workspace-server/platform.pid ]; then
+            kill "$(cat workspace-server/platform.pid)" 2>/dev/null || true
+          fi
+      - name: Stop service containers
+        if: always()
+        run: |
+          docker rm -f "$PG_CONTAINER" 2>/dev/null || true
+          docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true
+
+  # Real STAGING gate: provisions a throwaway org + sibling-per-runtime,
+  # drives the LITERAL list_peers MCP call per runtime, asserts 200 +
+  # expected peer set, then scoped teardown. push(main)/dispatch/cron only.
  peer-visibility:
    name: E2E Peer Visibility
    runs-on: ubuntu-latest
@@ -16,9 +16,9 @@ name: E2E Staging Canvas (Playwright)
 # e2e-staging-saas.yml (which tests the API shape) by exercising the
 # actual browser + canvas bundle against live staging.
 #
-# Triggers: push to main/staging or PR touching canvas sources + this workflow,
-# manual dispatch, and weekly cron to catch browser/runtime drift even
-# when canvas is quiet.
+# Triggers: push to main, PR touching canvas sources + this workflow only
+# after the PR enters `merge-queue`, manual dispatch, and scheduled cron to
+# catch browser/runtime drift even when canvas is quiet.
 # Added staging to push/pull_request branches so the auto-promote gate
 # check (--event push --branch staging) can see a completed run for this
 # workflow — mirrors what PR #1891 does for e2e-api.yml.
@@ -37,9 +37,10 @@ on:
  pull_request:
    branches: [main]
  schedule:
-    # Weekly on Sunday 08:00 UTC — catches Chrome / Playwright / Next.js
+    # Nightly at 08:00 UTC — catches Chrome / Playwright / Next.js
    # release-note-shaped regressions that don't ride in with a PR.
-    - cron: '0 8 * * 0'
+    - cron: '0 8 * * *'
+  workflow_dispatch:

 concurrency:
  # Per-SHA grouping (changed 2026-04-28 from a single global group). The
@@ -79,10 +80,13 @@ jobs:
        with:
          fetch-depth: 0
      - id: decide
+        env:
+          GITEA_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          QUEUE_LABEL: merge-queue
        # Inline replacement for dorny/paths-filter — see e2e-api.yml.
-        # Cron triggers always run real work (no diff context).
+        # Cron and manual triggers always run real work (no diff context).
        run: |
-          if [ "${{ github.event_name }}" = "schedule" ]; then
+          if [ "${{ github.event_name }}" = "schedule" ] || [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
            echo "canvas=true" >> "$GITHUB_OUTPUT"
            exit 0
          fi
@@ -102,9 +106,26 @@ jobs:
            exit 0
          fi
          CHANGED=$(git diff --name-only "$BASE" HEAD)
-          if echo "$CHANGED" | grep -qE '^(canvas/|\.gitea/workflows/e2e-staging-canvas\.yml$)'; then
+          if ! echo "$CHANGED" | grep -qE '^(canvas/|\.gitea/workflows/e2e-staging-canvas\.yml$)'; then
+            echo "canvas=false" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          if [ "${{ github.event_name }}" != "pull_request" ]; then
+            echo "canvas=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+
+          authfile=$(mktemp)
+          chmod 600 "$authfile"
+          printf 'header = "Authorization: token %s"\n' "$GITEA_TOKEN" > "$authfile"
+          labels=$(curl -fsS -K "$authfile" \
+            "${{ github.server_url }}/api/v1/repos/${{ github.repository }}/issues/${{ github.event.pull_request.number }}/labels" \
+            | python3 -c 'import json,sys; print("\n".join(label.get("name","") for label in json.load(sys.stdin)))')
+          rm -f "$authfile"
+          if printf '%s\n' "$labels" | grep -qx "$QUEUE_LABEL"; then
            echo "canvas=true" >> "$GITHUB_OUTPUT"
          else
+            echo "PR is not in merge-queue; skipping heavy E2E Staging Canvas for normal PR path."
            echo "canvas=false" >> "$GITHUB_OUTPUT"
          fi

@@ -169,7 +190,14 @@ jobs:
      - name: Install Playwright browsers
        if: needs.detect-changes.outputs.canvas == 'true'
        timeout-minutes: 10
-        run: npx playwright install --with-deps chromium
+        run: |
+          PREBAKED_PLAYWRIGHT=/ms-playwright
+          if [ -d "${PREBAKED_PLAYWRIGHT}" ] && find "${PREBAKED_PLAYWRIGHT}" -maxdepth 3 -type f -name 'chrome' | grep -q .; then
+            echo "Using prebaked Playwright Chromium from ${PREBAKED_PLAYWRIGHT}"
+            echo "PLAYWRIGHT_BROWSERS_PATH=${PREBAKED_PLAYWRIGHT}" >> "$GITHUB_ENV"
+            exit 0
+          fi
+          npx playwright install --with-deps chromium

      - name: Run staging canvas E2E
        if: needs.detect-changes.outputs.canvas == 'true'
@@ -13,8 +13,12 @@ name: gitea-merge-queue
 #   - add `merge-queue-hold` to pause a queued PR without removing it

 on:
-  schedule:
-    - cron: '*/5 * * * *'
+  # Schedule moved to operator-config:
+  #   /etc/cron.d/molecule-core-merge-queue ->
+  #   /usr/local/bin/molecule-core-cron-bot.sh merge-queue
+  #
+  # The queue bot still processes one PR per tick, but no longer occupies
+  # one of the shared Actions runners just to poll.
  workflow_dispatch:

 permissions:
@@ -52,5 +56,9 @@ jobs:
          # explicitly instead of the combined state avoids false-pause when
          # non-blocking jobs (continue-on-error: true) have failed — those
          # failures pollute combined state but do not gate merges.
+          # NOTE: the event-suffixed context name is intentional — branch protection
+          # MUST require `CI / all-required (pull_request)` (with suffix), NOT the
+          # bare `CI / all-required`. Gitea treats absent contexts as pending, not
+          # skipped; requiring the bare name silently blocks all merges (issue #1473).
          PUSH_REQUIRED_CONTEXTS: CI / all-required (push)
        run: python3 .gitea/scripts/gitea-merge-queue.py
@@ -77,7 +77,16 @@ env:
 jobs:
  detect-changes:
    name: detect-changes
-    runs-on: ubuntu-latest
+    # mc#1529 §1: pin to `docker-host` so the integration job runs on the
+    # operator-host runners (molecule-runner-*), which carry the
+    # `molecule-core-net` bridge network this workflow depends on. PC2
+    # runners (hongming-pc-runner-*) also advertise ubuntu-latest but
+    # don't have that network — the previous `runs-on: ubuntu-latest`
+    # rolled the dice and hard-failed the bridge-inspect step ~30% of
+    # the time. detect-changes itself doesn't need the bridge, but keeping
+    # both jobs on the same label avoids workspace-volume cross-host
+    # surprises and keeps the routing rule discoverable in one place.
+    runs-on: docker-host
    # mc#774 Phase 3 (RFC §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -129,7 +138,9 @@ jobs:
  integration:
    name: Handlers Postgres Integration
    needs: detect-changes
-    runs-on: ubuntu-latest
+    # mc#1529 §1: must run on operator-host (where `molecule-core-net`
+    # exists). See detect-changes for the full routing rationale.
+    runs-on: docker-host
    # mc#774 Phase 3 (RFC §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -62,7 +62,13 @@ env:
 jobs:
  # bp-exempt: change detector only; downstream Harness Replays is the meaningful gate.
  detect-changes:
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: pin to `docker-host` so this lane lands on
+    # Linux operator-host runners (the only ones with a working
+    # docker.sock + `molecule-core-net`). The bare `ubuntu-latest`
+    # label is also matched by hongming-pc-runner-* (Windows act_runner
+    # v1.0.3), where the `docker compose ...` exec below fails. Mirror
+    # of mc#1543; see internal#512 for class defect.
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -162,7 +168,9 @@ jobs:
  harness-replays:
    needs: detect-changes
    name: Harness Replays
-    runs-on: ubuntu-latest
+    # mc#1529 follow-on: `docker compose ... ps/logs` against tenant-alpha/
+    # beta containers. Must run on operator-host Linux (docker-host).
+    runs-on: docker-host
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -0,0 +1,168 @@
+name: Lint forbidden tenant-env keys
+
+# RFC#523 Layer 3 (task #146): scan workspace_secrets-writer Go code
+# under workspace-server/ for new code that hardcodes a forbidden
+# operator-scope env var NAME (GITEA_TOKEN, CP_ADMIN_API_TOKEN,
+# RAILWAY_TOKEN, INFISICAL_OPERATOR_TOKEN, MOLECULE_OPERATOR_*, …).
+#
+# Catches the class "a new writer accidentally widens the propagation
+# set" — e.g. a future env-mutator plugin that sets envVars["GITEA_TOKEN"]
+# directly. Today the L1 runtime guard would abort the provision, but
+# this lint surfaces the offending code at PR review time instead of
+# at first provision attempt.
+#
+# Companion layers:
+#   - L1: workspace-server/internal/handlers/workspace_provision_forbidden_env.go
+#         (fail-closed abort at provision time)
+#   - L2: workspace/entrypoint.sh top-of-file env-grep + exit 1
+#
+# Open-source-template-friendly: the deny pattern is generic. A fork
+# can copy this workflow and replace OPERATOR_KEY_PATTERN with its
+# own operator-scope key names.
+#
+# Path-filter discipline:
+#   This workflow runs on every PR (no paths: filter — see
+#   feedback_path_filtered_workflow_cant_be_required). The scan itself
+#   targets workspace_secrets-writer paths via grep -r; it's fast
+#   (sub-second) so unconditional run is fine.
+
+on:
+  pull_request:
+    types: [opened, synchronize, reopened]
+  push:
+    branches: [main, staging]
+
+env:
+  GITHUB_SERVER_URL: https://git.moleculesai.app
+
+jobs:
+  scan:
+    name: Scan workspace_secrets writers for forbidden env keys
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 1
+
+      - name: Scan for forbidden operator-scope env key NAMES in writer paths
+        run: |
+          set -euo pipefail
+
+          # Forbidden EXACT-MATCH env var names. Kept in lockstep with
+          # workspace-server/internal/handlers/workspace_provision_forbidden_env.go
+          # forbiddenTenantEnvKeys. The Go-side test
+          # TestIsForbiddenTenantEnvKey_ExactMatches is the source of
+          # truth — if Go-side adds a key, also add it here (and
+          # vice-versa). Drift between the two is the failure mode this
+          # entire 3-layer guardrail is designed to catch.
+          FORBIDDEN_KEYS=(
+            "GITEA_TOKEN" "GITEA_PAT"
+            "GITHUB_TOKEN" "GITHUB_PAT" "GH_TOKEN"
+            "GITLAB_TOKEN" "GL_TOKEN"
+            "BITBUCKET_TOKEN"
+            "CP_ADMIN_API_TOKEN" "CP_ADMIN_TOKEN"
+            "INFISICAL_OPERATOR_TOKEN" "INFISICAL_BOOTSTRAP_TOKEN"
+            "RAILWAY_TOKEN" "RAILWAY_PERSONAL_API_TOKEN"
+            "HETZNER_TOKEN" "HETZNER_API_TOKEN"
+          )
+
+          # Forbidden PREFIX patterns — operator-scope families.
+          FORBIDDEN_PREFIXES=(
+            "MOLECULE_OPERATOR_"
+          )
+
+          # Writer paths: Go source under workspace-server/ that
+          # writes to the env-vars map or to workspace_secrets DB rows.
+          # Tests, the forbidden-env source itself, and the silent-
+          # strip denylist are exempt (they LIST the keys by design).
+          SCAN_ROOT="workspace-server/internal"
+          # Exempt paths fall in two classes:
+          #   1. The deny-set definitions + the silent-strip denylist:
+          #      they LIST the forbidden names by design.
+          #   2. Pre-RFC#523 persona-merge / config-read paths that
+          #      already handle these names correctly (the silent-
+          #      strip downstream + the new L1 fail-closed cover the
+          #      runtime risk; these reads are unchanged).
+          # New code MUST NOT be added to this list without reviewer
+          # signoff and a one-line justification in this diff.
+          EXEMPT_PATHS=(
+            # Class 1 — deny-set definitions
+            "workspace-server/internal/handlers/workspace_provision_forbidden_env.go"
+            "workspace-server/internal/handlers/workspace_provision_forbidden_env_test.go"
+            "workspace-server/internal/provisioner/provisioner.go"
+            "workspace-server/internal/provisioner/provisioner_test.go"
+            # Class 2 — pre-existing persona-fallback / org-helper paths
+            # that set the GITEA_TOKEN fallback lane (stripped downstream
+            # by provisioner.buildContainerEnv per forensic #145). The
+            # new L1 fail-closed runs BEFORE these writers, so any
+            # operator-scope leak via global/workspace_secrets is
+            # already caught. See applyAgentGitHTTPCreds doc-comment.
+            "workspace-server/internal/handlers/agent_git_identity.go"
+            "workspace-server/internal/handlers/org_helpers.go"
+            "workspace-server/internal/handlers/org.go"
+            # Class 2 — CP→platform admin auth (NOT a tenant env write;
+            # this is the control-plane HTTP auth header source).
+            "workspace-server/internal/provisioner/cp_provisioner.go"
+          )
+
+          # Build a single grep -F pattern: every forbidden key wrapped
+          # in quotes (Go string-literal form, which is how env-map
+          # writes appear). e.g. envVars["GITEA_TOKEN"] = ... or
+          # `"GITEA_TOKEN":` in a literal-map declaration.
+          #
+          # We deliberately match the quoted form so a comment that
+          # happens to spell the name without quotes (e.g. "see
+          # GITEA_TOKEN below") doesn't trip the lint.
+          PATTERN=""
+          for k in "${FORBIDDEN_KEYS[@]}"; do
+            PATTERN="${PATTERN}\"${k}\"\n"
+          done
+          for p in "${FORBIDDEN_PREFIXES[@]}"; do
+            # Prefix match needs a regex; switch to grep -E below for
+            # this slice. Kept conceptually here so the deny set lives
+            # in one place; scan is run twice (literal + prefix).
+            true
+          done
+
+          # Build exempt-paths grep filter — `grep -v -f` style.
+          EXEMPT_FILTER=$(mktemp)
+          trap 'rm -f "$EXEMPT_FILTER"' EXIT
+          for p in "${EXEMPT_PATHS[@]}"; do
+            echo "$p" >> "$EXEMPT_FILTER"
+          done
+
+          # --- Exact-match scan ---
+          HITS=""
+          for k in "${FORBIDDEN_KEYS[@]}"; do
+            # Only .go files; skip _test.go for the writer-path scan
+            # since tests legitimately reference the names. The
+            # writer-path lint targets PRODUCTION code only.
+            found=$(grep -rn --include='*.go' --exclude='*_test.go' "\"${k}\"" "$SCAN_ROOT" 2>/dev/null \
+                    | grep -v -F -f "$EXEMPT_FILTER" || true)
+            if [ -n "$found" ]; then
+              HITS="${HITS}${found}\n"
+            fi
+          done
+
+          # --- Prefix scan ---
+          for prefix in "${FORBIDDEN_PREFIXES[@]}"; do
+            found=$(grep -rnE --include='*.go' --exclude='*_test.go' "\"${prefix}[A-Z0-9_]+\"" "$SCAN_ROOT" 2>/dev/null \
+                    | grep -v -F -f "$EXEMPT_FILTER" || true)
+            if [ -n "$found" ]; then
+              HITS="${HITS}${found}\n"
+            fi
+          done
+
+          if [ -n "$HITS" ]; then
+            echo "::error::RFC#523 Layer 3: forbidden operator-scope env var name(s) hardcoded in tenant-workspace writer paths:"
+            printf "$HITS"
+            echo ""
+            echo "These env-var NAMES are on the operator-scope deny list (see"
+            echo "workspace-server/internal/handlers/workspace_provision_forbidden_env.go)."
+            echo "If your code legitimately needs to inject one of these for a"
+            echo "non-tenant code path, add the file to EXEMPT_PATHS in this"
+            echo "workflow with a one-line justification — reviewer signoff required."
+            exit 1
+          fi
+
+          echo "OK No forbidden operator-scope env key names hardcoded in writer paths."
@@ -0,0 +1,182 @@
+name: Lint no tenant GITEA or GITHUB token write
+
+# Task #146 — CI guardrail companion to RFC#523's `lint-forbidden-env-keys.yml`.
+#
+# `lint-forbidden-env-keys.yml` (Layer 3) catches code that hardcodes a
+# forbidden env-var key NAME as a quoted literal in workspace_secrets
+# writer paths under workspace-server/internal/.
+#
+# This workflow catches a BROADER class: any code path that reads a
+# repo-host token (GITEA_TOKEN / GITHUB_TOKEN / GH_TOKEN) and then writes
+# it into a TENANT WORKSPACE's env, secret store, user-data, or
+# provision payload. This is the actual RFC#523 threat-model statement —
+# the goal is "no tenant workspace ever receives an operator-scope repo
+# token," not just "no _quoted_ literal `GITEA_TOKEN`." A future writer
+# could route the value via a variable, a struct field, or a config key
+# and slip past the existing literal scan; this lint catches those
+# routing patterns at PR review time.
+#
+# Scope
+#   Scans the WHOLE repo's Go sources (not just workspace-server/) for
+#   co-occurrences of:
+#     - a repo-host token NAME (GITEA_TOKEN / GITHUB_TOKEN / GH_TOKEN /
+#       GITEA_PAT / GITHUB_PAT) used as os.Getenv argument or string
+#       literal
+#     - within a file that ALSO references a tenant-writer surface
+#       (`tenant`, `workspace_secrets`, `global_secrets`, `seedAllowList`,
+#       `/settings/secrets`, `userData`, `provisionPayload`,
+#       `envVars[`, `containerEnv`).
+#
+#   Co-occurrence (not single-line) is the false-positive control: a
+#   file that just LOGS the variable name (e.g. "missing GITEA_TOKEN")
+#   without touching any tenant surface won't fire.
+#
+# Drift contract with lint-forbidden-env-keys.yml
+#   Both lints share the same FORBIDDEN_KEYS list (a subset — only the
+#   repo-host tokens, since this lint's threat model is "tenant gets
+#   write access to operator's git host"). If RFC#523's deny set grows,
+#   update BOTH this file AND lint-forbidden-env-keys.yml AND the Go
+#   source-of-truth in
+#   workspace-server/internal/handlers/workspace_provision_forbidden_env.go.
+#
+# Open-source-template-friendly
+#   The patterns scanned are generic (no MOLECULE_-prefix literals).
+#   A fork can copy this workflow as-is and adjust FORBIDDEN_KEYS.
+#
+# Path-filter discipline
+#   No `paths:` filter — required-status workflows must run on every PR
+#   per `feedback_path_filtered_workflow_cant_be_required`. Scan is
+#   sub-second.
+
+on:
+  pull_request:
+    types: [opened, synchronize, reopened]
+  push:
+    branches: [main, staging]
+
+env:
+  GITHUB_SERVER_URL: https://git.moleculesai.app
+
+jobs:
+  # bp-exempt: advisory RFC#523 lint; PR review gate is review-driven, not BP-driven.
+  # (Carried with the workflow-name rename in PR mc#1593 so the renamed
+  # context emission satisfies lint_required_context_exists_in_bp Tier 2g.)
+  scan:
+    name: Scan for repo-host token write into tenant workspace surface
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 1
+
+      - name: Find Go files referencing a tenant-writer surface AND a repo-host token
+        run: |
+          set -euo pipefail
+
+          # Repo-host token NAMES — the threat-model subset. Operator-fleet
+          # tokens (CP_ADMIN_API_TOKEN, RAILWAY_TOKEN, INFISICAL_*) are
+          # caught by lint-forbidden-env-keys.yml's broader deny set; this
+          # lint focuses on the git-host class so a single co-occurrence
+          # match has a low false-positive rate.
+          FORBIDDEN_KEYS=(
+            "GITEA_TOKEN"
+            "GITEA_PAT"
+            "GITHUB_TOKEN"
+            "GITHUB_PAT"
+            "GH_TOKEN"
+          )
+
+          # Tenant-writer surface markers. A file matches the surface set
+          # if it references ANY of these strings. This is the "is this
+          # code path writing into a tenant workspace?" heuristic.
+          # Curated to catch the actual code shapes used in this repo
+          # (verified by grep against current main 2026-05-19):
+          #   - "workspace_secrets" / "global_secrets"  → DB table writes
+          #   - "seedAllowList"                          → CP-side seed table
+          #   - "/settings/secrets"                      → tenant HTTP API write
+          #   - "envVars["                               → in-memory env map write
+          #   - "containerEnv"                           → docker-run env-set
+          #   - "userData"                               → EC2 user-data script
+          #   - "provisionPayload" / "provisionContext"  → provision-request shape
+          SURFACE_PATTERN='workspace_secrets|global_secrets|seedAllowList|/settings/secrets|envVars\[|containerEnv|userData|provisionPayload|provisionContext'
+
+          # Files that legitimately reference these names AND a surface
+          # marker, but do so for guard / strip / test / doc-comment
+          # reasons. New entries require reviewer signoff and a one-line
+          # justification in the diff.
+          EXEMPT_FILES=(
+            # RFC#523 L1 deny-set source-of-truth + tests
+            "workspace-server/internal/handlers/workspace_provision_forbidden_env.go"
+            "workspace-server/internal/handlers/workspace_provision_forbidden_env_test.go"
+            # Forensic-#145 silent-strip denylist (defense-in-depth, by design lists the names)
+            "workspace-server/internal/provisioner/provisioner.go"
+            "workspace-server/internal/provisioner/provisioner_test.go"
+            # Pre-RFC#523 persona-fallback / org-helper paths. The L1
+            # fail-closed runs BEFORE these writers; downstream silent-strip
+            # also covers them. See applyAgentGitHTTPCreds doc-comment.
+            "workspace-server/internal/handlers/agent_git_identity.go"
+            "workspace-server/internal/handlers/org_helpers.go"
+            "workspace-server/internal/handlers/org.go"
+            # CP→platform admin auth (NOT a tenant env write).
+            "workspace-server/internal/provisioner/cp_provisioner.go"
+          )
+
+          # Build an extended-regex alternation of forbidden keys.
+          KEY_ALT="$(IFS='|'; echo "${FORBIDDEN_KEYS[*]}")"
+
+          # Find candidate files: Go non-test sources that contain a
+          # tenant-writer surface marker.
+          mapfile -t CANDIDATES < <(
+            grep -rlE --include='*.go' --exclude='*_test.go' \
+              "${SURFACE_PATTERN}" . 2>/dev/null \
+            | sed 's|^\./||' \
+            | sort -u
+          )
+
+          if [ "${#CANDIDATES[@]}" -eq 0 ]; then
+            echo "OK No tenant-writer-surface files found in tree (unexpected, but not a lint failure)."
+            exit 0
+          fi
+
+          HITS=""
+          for f in "${CANDIDATES[@]}"; do
+            # Skip exempt files.
+            skip=0
+            for ex in "${EXEMPT_FILES[@]}"; do
+              if [ "$f" = "$ex" ]; then skip=1; break; fi
+            done
+            [ "$skip" = "1" ] && continue
+
+            # File contains a surface marker; now grep for a forbidden
+            # key NAME. We require a QUOTED-literal match to avoid
+            # firing on a comment like "// also handle GITEA_TOKEN".
+            #
+            # The literal form catches:
+            #   - os.Getenv("GITEA_TOKEN")
+            #   - envVars["GITEA_TOKEN"] = ...
+            #   - {envKey: "GITEA_TOKEN", tenantKey: "GITEA_TOKEN"}
+            # but not:
+            #   - // see GITEA_TOKEN below   (no quotes)
+            found=$(grep -nE "\"(${KEY_ALT})\"" "$f" 2>/dev/null || true)
+            if [ -n "$found" ]; then
+              HITS="${HITS}--- ${f} ---\n${found}\n"
+            fi
+          done
+
+          if [ -n "$HITS" ]; then
+            echo "::error::Task #146 lint: repo-host token name(s) quoted in a tenant-writer-surface file:"
+            printf "$HITS"
+            echo ""
+            echo "These files reference a tenant-writer surface (workspace_secrets,"
+            echo "seedAllowList, /settings/secrets, containerEnv, userData, etc.)"
+            echo "AND quote a repo-host token name (GITEA_TOKEN/GITHUB_TOKEN/…)."
+            echo "Per RFC#523 threat model, tenant workspaces MUST NOT receive"
+            echo "operator-scope repo-host tokens. If your code legitimately needs"
+            echo "to reference one of these names in a tenant-writer file (e.g."
+            echo "a deny-set definition or silent-strip list), add the file to"
+            echo "EXEMPT_FILES with a one-line justification — reviewer signoff"
+            echo "required."
+            exit 1
+          fi
+
+          echo "OK No tenant-writer-surface file co-mentions a repo-host token literal."
@@ -0,0 +1,164 @@
+name: lint-required-workflows-docker-host-pinned
+
+# Fail-closed lint that catches workflows touching docker.sock without
+# pinning `runs-on:` to a Linux-only label.
+#
+# Class defect (internal#512 + mc#1529 + today's oc#81/82/83 + autogen#8):
+# the `ubuntu-latest` label is advertised by BOTH the Linux operator-host
+# runners (molecule-runner-*) AND the Windows act_runner v1.0.3 on
+# hongming-pc-runner-*. Job placement is non-deterministic. When a docker-
+# bound job lands on a Windows runner, `docker run`/`docker login`/
+# `docker compose` fail with platform-specific errors ("protocol not
+# available", "cannot exec", etc.) — placement-dependent, not transient.
+#
+# This lint enforces the convention: any workflow whose YAML body
+# contains a docker exec (`docker run|build|buildx|compose|pull|push|
+# exec|tag|login|cp|inspect|ps` OR `docker/build-push-action|docker/
+# login-action|docker/setup-buildx`) MUST pin every job's `runs-on:` to
+# one of:
+#   - docker-host  (general docker.sock work — molecule-runner-*)
+#   - publish      (image build/push — molecule-runner-publish-*)
+#
+# Comments and heredoc/markdown bodies that merely MENTION docker are
+# excluded by the detection rule (see scan.py below).
+#
+# Per `feedback_never_skip_ci`: this is fail-closed (exit 1 on miss).
+
+on:
+  pull_request:
+    paths:
+      - '.gitea/workflows/**'
+  push:
+    branches: [main, staging]
+    paths:
+      - '.gitea/workflows/**'
+
+permissions:
+  contents: read
+
+env:
+  GITHUB_SERVER_URL: https://git.moleculesai.app
+
+jobs:
+  lint-docker-host-pin:
+    name: Lint docker-host pin on docker-touching workflows
+    runs-on: docker-host
+    timeout-minutes: 5
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Scan workflows for docker-bound jobs missing docker-host/publish pin
+        run: |
+          set -euo pipefail
+          python3 - <<'PY'
+          import os
+          import re
+          import sys
+
+          # Docker-step detection: real exec, not just word-mention in comments.
+          # We strip comment-only lines, then look for the docker subcommand
+          # tokens at word-boundary, OR uses: docker/* actions.
+          DOCKER_EXEC = re.compile(
+              r'(?<!\w)docker\s+(run|build|buildx|compose|pull|push|exec|tag|login|cp|inspect|ps)\b'
+          )
+          DOCKER_ACTION = re.compile(
+              r'uses:\s*docker/(build-push-action|login-action|setup-buildx-action|setup-qemu-action)'
+          )
+          # Detect a job header line like `  myjob:` (2-space indent) AND its runs-on.
+          JOB_HEADER = re.compile(r'^( {2})([a-zA-Z0-9_-]+):\s*$')
+          RUNS_ON    = re.compile(r'^( {4})runs-on:\s*(.+?)\s*$')
+
+          ALLOWED_LABELS = {'docker-host', 'publish'}
+
+          fails = []
+          warnings = []
+
+          # Gitea is SSOT for molecule-core CI per task #347 / memory
+          # reference_molecule_core_actions_gitea_only. The legacy
+          # .github/workflows/ tree was deleted in SSOT-Instance-4 (#331).
+          roots = []
+          for root in ('.gitea/workflows',):
+              if os.path.isdir(root):
+                  roots.append(root)
+
+          for root in roots:
+              for fn in sorted(os.listdir(root)):
+                  if not (fn.endswith('.yml') or fn.endswith('.yaml')):
+                      continue
+                  path = os.path.join(root, fn)
+                  with open(path) as f:
+                      raw_lines = f.readlines()
+
+                  # Parse job headers + their runs-on. Simple line scan; relies on
+                  # 2-space job indent + 4-space runs-on indent under `jobs:`.
+                  jobs = []
+                  current = None
+                  in_jobs = False
+                  for i, line in enumerate(raw_lines, 1):
+                      if re.match(r'^jobs:\s*$', line):
+                          in_jobs = True
+                          continue
+                      if not in_jobs:
+                          continue
+                      mh = JOB_HEADER.match(line)
+                      if mh:
+                          if current:
+                              current['end'] = i - 1
+                              jobs.append(current)
+                          current = {'name': mh.group(2), 'line': i, 'end': len(raw_lines), 'runs_on': None}
+                          continue
+                      mr = RUNS_ON.match(line)
+                      if mr and current and current['runs_on'] is None:
+                          current['runs_on'] = mr.group(2).strip()
+                  if current:
+                      jobs.append(current)
+
+                  for j in jobs:
+                      # Strip pure-comment lines for docker-exec detection so
+                      # documentation comments don't trigger the lint. Scan the
+                      # current job body only: a workflow may contain one
+                      # docker-bound job and several harmless metadata jobs.
+                      job_lines = raw_lines[j['line'] - 1:j['end']]
+                      scan_text = ''.join(
+                          l for l in job_lines
+                          if not re.match(r'^\s*#', l)
+                      )
+                      has_docker = bool(DOCKER_EXEC.search(scan_text)) or bool(DOCKER_ACTION.search(scan_text))
+                      if not has_docker:
+                          continue
+                      ro = j['runs_on']
+                      if ro is None:
+                          # Reusable workflow caller (`uses:` instead of `runs-on:`) —
+                          # skip; rule enforced in the called workflow.
+                          continue
+                      # Strip surrounding [ ] and quotes.
+                      ro_norm = ro.strip('[]').strip().strip('"\'')
+                      # Multi-label "[a, b]" — split.
+                      labels = [t.strip().strip('"\'') for t in ro_norm.split(',') if t.strip()]
+                      if any(lbl in ALLOWED_LABELS for lbl in labels):
+                          continue
+                      # Allow caller-supplied label expressions; spell the
+                      # marker indirectly so Gitea's expression parser does
+                      # not try to parse this Python heredoc.
+                      expression_marker = '$' + '{{'
+                      if any(expression_marker in lbl for lbl in labels):
+                          continue
+                      fails.append(
+                          f"{path}:{j['line']}: job `{j['name']}` uses docker but runs-on={ro!r} "
+                          f"(must be one of {sorted(ALLOWED_LABELS)})"
+                      )
+
+          if fails:
+              print("FAIL: docker-bound jobs missing docker-host/publish pin:")
+              for f in fails:
+                  print(f"  - {f}")
+              print()
+              print("Why this rule exists (internal#512 + mc#1529):")
+              print("  Bare `ubuntu-latest` is advertised by BOTH Linux operator-host")
+              print("  runners AND Windows hongming-pc-runner-* (act_runner v1.0.3).")
+              print("  Docker-bound jobs that land on Windows fail non-deterministically.")
+              print("  Pin to `docker-host` (general) or `publish` (image build/push).")
+              sys.exit(1)
+
+          print("OK: all docker-bound jobs are pinned to docker-host or publish.")
+          PY
@@ -0,0 +1,88 @@
+name: Lint shellcheck (arm64 pilot)
+
+# Mac-CI dual-track pilot (#233). ADDITIVE / NOT REQUIRED.
+#
+# Validates the arm64 self-hosted lane (no docker.sock, no privileged
+# ops) before any required gate moves onto it. Until a Mac arm64 runner
+# is registered with the `arm64` label, this workflow sits PENDING —
+# that is FINE: `arm64` is NOT in branch_protections required contexts.
+#
+# Pairs with internal#543 (RFC: Mac arm64 multi-arch runner-base).
+# No paths: filter on purpose (feedback_path_filtered_workflow_cant_be_required).
+
+on:
+  pull_request:
+    branches:
+      - main
+      - staging
+  push:
+    branches:
+      - main
+
+permissions:
+  contents: read
+
+jobs:
+  shellcheck-arm64:
+    name: shellcheck-arm64 (pilot)
+    runs-on: [self-hosted, arm64]
+    # NOT a required check; safe to sit pending until Mac runner is up.
+    # If the Mac runner has trouble pulling actions/checkout we fall
+    # back to a plain git clone (see step 'fallback clone').
+    timeout-minutes: 10
+    env:
+      GITHUB_SERVER_URL: https://git.moleculesai.app
+    steps:
+      - name: Identify runner
+        run: |
+          set -eu
+          echo "arch=$(uname -m)"
+          echo "kernel=$(uname -sr)"
+          echo "shell=$BASH_VERSION"
+          # Sanity: must actually be arm64. If amd64 sneaks in here,
+          # fail fast — that means the label routing is wrong.
+          case "$(uname -m)" in
+            aarch64|arm64) echo "arm64 confirmed" ;;
+            *) echo "ERROR: expected arm64, got $(uname -m)"; exit 1 ;;
+          esac
+
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+
+      - name: Install shellcheck (arm64)
+        run: |
+          set -eu
+          if command -v shellcheck >/dev/null 2>&1; then
+            echo "shellcheck already present: $(shellcheck --version | head -1)"
+          else
+            # Prefer apt if the runner base ships it; else download arm64 binary.
+            if command -v apt-get >/dev/null 2>&1; then
+              sudo apt-get update -qq
+              sudo apt-get install -y --no-install-recommends shellcheck
+            else
+              SC_VER=v0.10.0
+              curl -fsSL "https://github.com/koalaman/shellcheck/releases/download/${SC_VER}/shellcheck-${SC_VER}.linux.aarch64.tar.xz" \
+                | tar -xJf - --strip-components=1
+              sudo mv shellcheck /usr/local/bin/
+            fi
+          fi
+          shellcheck --version | head -2
+
+      - name: Run shellcheck on .gitea/scripts/*.sh
+        run: |
+          set -eu
+          # Only the scripts we control under .gitea/scripts. Pilot
+          # scope is intentionally narrow — broaden in a follow-up
+          # once the lane is proven.
+          mapfile -t TARGETS < <(find .gitea/scripts -maxdepth 2 -type f -name '*.sh' | sort)
+          if [ "${#TARGETS[@]}" -eq 0 ]; then
+            echo "No .sh files found under .gitea/scripts — nothing to check"
+            exit 0
+          fi
+          echo "Checking ${#TARGETS[@]} file(s):"
+          printf '  %s\n' "${TARGETS[@]}"
+          # SC1091 = couldn't follow non-constant source; expected for
+          # CI-time analysis without the full runtime layout.
+          shellcheck --severity=error --exclude=SC1091 "${TARGETS[@]}"
@@ -42,20 +42,30 @@ permissions:
  packages: write

 env:
-  IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/canvas
+  # SSOT-Instance-10 (#333): ECR registry triplet (account.dkr.ecr.region.amazonaws.com)
+  # sourced from org/repo var `ECR_REGISTRY` with the current prod-account literal as
+  # bootstrap fallback. When the org var is set, the fallback becomes dead code and
+  # switching accounts/regions is a one-line change at the org level (instead of
+  # touching every workflow). Pattern mirrors `vars.CP_URL || 'literal'` already in
+  # use below in this repo's staging-verify.yml.
+  IMAGE_NAME: ${{ vars.ECR_REGISTRY || '153263036946.dkr.ecr.us-east-2.amazonaws.com' }}/molecule-ai/canvas
  GITHUB_SERVER_URL: https://git.moleculesai.app

 jobs:
  # bp-exempt: post-merge image publication side effect; CI / all-required gates source changes.
  build-and-push:
    name: Build & push canvas image
-    # REVERTED (infra/revert-docker-runner-label): `runs-on: ubuntu-latest` restored.
-    # The `docker` label is not registered on any act_runner. `runs-on: [ubuntu-latest, docker]`
-    # causes jobs to queue indefinitely with zero eligible runners — strictly worse than the
-    # pre-#599 coin-flip (50% success rate). Once the `docker` label is registered on
-    # ≥2 runners, re-apply the fix from #599 (infra/docker-runner-label).
-    # See issue #576 + infra-lead pulse ~00:30Z.
-    runs-on: ubuntu-latest
+    # Dedicated publish/release lane (internal#462 / #394 / #399). Ship
+    # path (on: push:main, canvas/**) — reserved capacity so a merged
+    # canvas fix's image build never FIFO-queues behind PR required-CI.
+    # The `publish` label resolves ONLY to the molecule-runner-publish-*
+    # sub-pool (config.publish.yaml). HARD DEPENDENCY: this MUST land
+    # AFTER the publish-lane runners are registered/advertising `publish`
+    # — the earlier #599 `docker` label attempt queued indefinitely with
+    # zero eligible runners precisely because the label was targeted
+    # before any runner advertised it (see #576). The lane is registered
+    # in this rollout (internal#462) so the precondition holds.
+    runs-on: publish
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -29,6 +29,14 @@ on:
  pull_request:
    paths:
      - "workspace/**"
+      # mc#1578 / a05add29 cure: build_runtime_package.py owns PYPROJECT_TEMPLATE
+      # (deps, classifiers, project metadata). A change there is publish-affecting
+      # even when workspace/** is untouched, so the autobump must fire to claim
+      # the next runtime-v$VERSION tag. Without this, manual tagging races PyPI
+      # (e.g. runtime-v0.1.18 collided with the 2026-04-27 PyPI 0.1.18 publish,
+      # blocking the python-multipart pin from reaching prod).
+      - "scripts/build_runtime_package.py"
+      - "scripts/test_build_runtime_package.py"
  # Bump-and-tag on main/staging push (the actual operational trigger).
  push:
    branches:
@@ -36,6 +44,8 @@ on:
      - staging
    paths:
      - "workspace/**"
+      - "scripts/build_runtime_package.py"
+      - "scripts/test_build_runtime_package.py"
  # Manual dispatch — useful when Gitea Actions API (/actions/*) is
  # unreachable (e.g. act_runner 404 on Gitea 1.22.6) and we cannot
  # re-trigger via curl.
@@ -104,7 +114,7 @@ jobs:
        with:
          python-version: "3.11"

-      - name: Compute next version from PyPI latest
+      - name: Compute next version from PyPI latest and existing tags
        id: bump
        run: |
          set -eu
@@ -112,9 +122,24 @@ jobs:
            | python -c "import sys,json; print(json.load(sys.stdin)['info']['version'])")
          MAJOR=$(echo "$LATEST" | cut -d. -f1)
          MINOR=$(echo "$LATEST" | cut -d. -f2)
-          PATCH=$(echo "$LATEST" | cut -d. -f3)
-          VERSION="${MAJOR}.${MINOR}.$((PATCH+1))"
-          echo "PyPI latest=$LATEST -> next=$VERSION"
+          TAG_LATEST=$(git tag --list "runtime-v${MAJOR}.${MINOR}.*" \
+            | sed -E 's/^runtime-v//' \
+            | grep -E '^[0-9]+\.[0-9]+\.[0-9]+$' \
+            | sort -V \
+            | tail -1 || true)
+          VERSION=$(PYPI_LATEST="$LATEST" TAG_LATEST="$TAG_LATEST" python - <<'PY'
+          import os
+
+          def parse(v):
+              return tuple(int(part) for part in v.split("."))
+
+          pypi = os.environ["PYPI_LATEST"]
+          tag = os.environ.get("TAG_LATEST") or pypi
+          base = max(parse(pypi), parse(tag))
+          print(f"{base[0]}.{base[1]}.{base[2] + 1}")
+          PY
+          )
+          echo "PyPI latest=$LATEST, latest runtime tag=${TAG_LATEST:-none} -> next=$VERSION"
          if ! echo "$VERSION" | grep -qE '^[0-9]+\.[0-9]+\.[0-9]+$'; then
            echo "::error::computed version $VERSION does not match PEP 440 X.Y.Z"
            exit 1
@@ -66,7 +66,10 @@ concurrency:

 jobs:
  publish:
-    runs-on: ubuntu-latest
+    # Dedicated publish/release lane (internal#462 / #394 / #399). Ship
+    # path (on: push tag runtime-v*) — reserved capacity, never FIFO
+    # behind PR-CI. `publish` resolves only to molecule-runner-publish-*.
+    runs-on: publish
    outputs:
      version: ${{ steps.version.outputs.version }}
      wheel_sha256: ${{ steps.wheel_hash.outputs.wheel_sha256 }}
@@ -138,7 +141,59 @@ jobs:
          /tmp/smoke/bin/pip install --quiet dist/*.whl
          /tmp/smoke/bin/python "$GITHUB_WORKSPACE/scripts/wheel_smoke.py"

-      - name: Publish to PyPI
+      # ─────────────────────────────────────────────────────────────────────
+      # RFC#596 (2026-05-19): Gitea PyPI registry as PRIMARY, PyPI as
+      # best-effort fallback. Eliminates the SPOF that caused the
+      # 2026-05-19 P0 (PyPI abuse-block #593 + Railway outage #595).
+      #
+      # Order is inverted intentionally:
+      #   1. Gitea FIRST — must succeed (our internal SSOT).
+      #   2. PyPI SECOND — best-effort, non-fatal on failure (courtesy
+      #      mirror; our consumers don't depend on it after Phase 4
+      #      template Dockerfile updates).
+      #
+      # Endpoint shape (verified live in RFC#596 Phase 5):
+      #   POST https://git.moleculesai.app/api/packages/molecule-ai/pypi/
+      #   HTTP Basic auth: username = gitea username, password = PAT with
+      #   `write:package` scope. Returns 201 Created on success.
+      # ─────────────────────────────────────────────────────────────────────
+
+      - name: Publish to Gitea PyPI registry (PRIMARY)
+        id: gitea_publish
+        working-directory: ${{ runner.temp }}/runtime-build
+        env:
+          # MOLECULE_PYPI_GITEA_PUBLISHER_USER: Gitea username for the publisher
+          # persona (must own a token with `write:package` scope).
+          # Provisioned in RFC#596 Phase 3 (operator-config PR).
+          # NOTE: secret name MUST NOT start with `GITEA_` or `GITHUB_` —
+          # Gitea 1.22.6 reserves those prefixes for built-in env vars and
+          # rejects repo-secret PUT with HTTP 400 / "invalid secret name".
+          # Empirically reproduced 2026-05-19 against
+          # `/repos/molecule-ai/molecule-core/actions/secrets/GITEA_*`.
+          MOLECULE_PYPI_GITEA_PUBLISHER_USER: ${{ secrets.MOLECULE_PYPI_GITEA_PUBLISHER_USER }}
+          # MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN: PAT for the publisher persona,
+          # `write:package` scope on molecule-ai org.
+          # Synced from Infisical /ci/gitea-pypi-publisher (RFC#596 Phase 3).
+          MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN: ${{ secrets.MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN }}
+        run: |
+          set -eu
+          if [ -z "${MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN:-}" ] || [ -z "${MOLECULE_PYPI_GITEA_PUBLISHER_USER:-}" ]; then
+            echo "::error::MOLECULE_PYPI_GITEA_PUBLISHER_USER / MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN secrets are not set."
+            echo "::error::Provision them via the RFC#596 Phase 3 operator-config sync script."
+            echo "::error::Gitea is the PRIMARY index per RFC#596 — publish job aborts here, NOT after PyPI."
+            exit 1
+          fi
+          python -m twine upload \
+            --verbose \
+            --repository-url "https://git.moleculesai.app/api/packages/molecule-ai/pypi/" \
+            --username "$MOLECULE_PYPI_GITEA_PUBLISHER_USER" \
+            --password "$MOLECULE_PYPI_GITEA_PUBLISHER_TOKEN" \
+            dist/*
+          echo "gitea_status=success" >> "$GITHUB_OUTPUT"
+          echo "gitea_url=https://git.moleculesai.app/api/packages/molecule-ai/pypi/simple/molecule-ai-workspace-runtime" >> "$GITHUB_OUTPUT"
+
+      - name: Publish to PyPI (FALLBACK, best-effort)
+        id: pypi_publish
        # working-directory matches the preceding Build/Verify steps. Without
        # this, twine runs from the default workspace checkout dir where
        # `dist/` doesn't exist and fails with:
@@ -146,6 +201,11 @@ jobs:
        # Caught on the first-ever successful dispatch of this workflow
        # (run 5097, 2026-05-11 02:08Z) — every other step in the publish
        # job already had this working-directory; Publish was missing it.
+        #
+        # RFC#596: this step is `continue-on-error: true` because PyPI is
+        # NO LONGER the primary index. PyPI 403/timeout/abuse-block does
+        # NOT block the publish — Gitea already has the wheel.
+        continue-on-error: true
        working-directory: ${{ runner.temp }}/runtime-build
        env:
          # PYPI_TOKEN: repository secret scoped to molecule-ai-workspace-runtime.
@@ -154,19 +214,48 @@ jobs:
          PYPI_TOKEN: ${{ secrets.PYPI_TOKEN }}
        run: |
          if [ -z "$PYPI_TOKEN" ]; then
-            echo "::error::PYPI_TOKEN secret is not set — set it at Settings → Actions → Variables and Secrets → New Secret."
-            echo "::error::Required format: pypi-XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
-            exit 1
+            echo "::warning::PYPI_TOKEN secret is not set — skipping PyPI mirror publish (non-fatal per RFC#596)."
+            echo "pypi_status=skipped_no_token" >> "$GITHUB_OUTPUT"
+            exit 0
          fi
-          python -m twine upload \
+          if python -m twine upload \
+            --verbose \
            --repository pypi \
            --username __token__ \
            --password "$PYPI_TOKEN" \
-            dist/*
+            dist/*; then
+            echo "pypi_status=success" >> "$GITHUB_OUTPUT"
+          else
+            rc=$?
+            echo "::warning::PyPI mirror publish failed (exit $rc). Non-fatal per RFC#596 — Gitea has the wheel."
+            echo "pypi_status=failed_exit_$rc" >> "$GITHUB_OUTPUT"
+          fi
+          echo "pypi_url=https://pypi.org/project/molecule-ai-workspace-runtime/${{ steps.version.outputs.version }}/" >> "$GITHUB_OUTPUT"
+
+      - name: Publish job summary (Gitea + PyPI status)
+        if: always()
+        run: |
+          {
+            echo "## publish-runtime $(date -u +%FT%TZ)"
+            echo
+            echo "**Version:** \`${{ steps.version.outputs.version }}\`"
+            echo "**Wheel SHA256:** \`${{ steps.wheel_hash.outputs.wheel_sha256 }}\`"
+            echo
+            echo "### Indexes"
+            echo
+            echo "| Index   | Status                                          | URL |"
+            echo "|---------|-------------------------------------------------|-----|"
+            echo "| Gitea (PRIMARY) | ${{ steps.gitea_publish.outputs.gitea_status || 'failed' }} | ${{ steps.gitea_publish.outputs.gitea_url || '—' }} |"
+            echo "| PyPI (fallback) | ${{ steps.pypi_publish.outputs.pypi_status || 'failed' }}  | ${{ steps.pypi_publish.outputs.pypi_url || '—' }} |"
+            echo
+            echo "Per RFC#596: Gitea is the contract. PyPI is best-effort."
+          } >> "$GITHUB_STEP_SUMMARY"

  cascade:
    needs: publish
-    runs-on: ubuntu-latest
+    # Publish/release lane (internal#462) — downstream of the runtime
+    # publish ship job; keep it on the reserved lane too.
+    runs-on: publish
    steps:
      - name: Wait for PyPI to propagate the new version
        env:
@@ -178,6 +267,12 @@ jobs:
            echo "::error::publish job did not expose wheel_sha256 — cannot verify wheel content. Refusing to fan out cascade."
            exit 1
          fi
+          # NOTE (RFC#596 follow-up): this propagation probe still resolves
+          # against PyPI's default index. After RFC#596 Phase 4 lands and
+          # consumers pull from Gitea first, this probe should be rewritten
+          # to verify the Gitea simple/ endpoint serves the new wheel
+          # (PyPI may be best-effort-failed and the cascade should still
+          # fan out, since templates will pull from Gitea). Tracked in #596.
          python -m venv /tmp/propagation-probe
          PROBE=/tmp/propagation-probe/bin
          $PROBE/pip install --upgrade --quiet pip
@@ -261,7 +356,10 @@ jobs:
          fi

          GITEA_URL="${GITEA_URL:-https://git.moleculesai.app}"
-          TEMPLATES="claude-code hermes openclaw codex langgraph crewai autogen deepagents gemini-cli"
+          # Keep in lockstep with manifest.json workspace_templates (suffix-stripped).
+          # Guarded by scripts/check-cascade-list-vs-manifest.sh (cascade-list-drift-gate).
+          # 2026-05-19: pruned crewai/deepagents/gemini-cli — not in manifest.
+          TEMPLATES="claude-code hermes openclaw codex langgraph autogen"
          FAILED=""
          SKIPPED=""

@@ -43,18 +43,39 @@ on:
 # `cancel-in-progress: false`; that is not acceptable for a workflow with a
 # production deploy job. Per-SHA image tags are immutable, and staging-latest is
 # best-effort last-writer-wins metadata.
+#
+# 2026-05-20 retrigger: run #86994 on mc#1589 merge sha 0f0f1ba2 failed at
+# setup-buildx-action with EACCES on PC2 WSL publish runner — the runner's
+# DOCKER_CONFIG=/home/hongming/.docker-ecr/ dir didn't have a buildx/certs
+# subdir writable by the container's UID 1001. Hot-patched the dir perms;
+# this chore push retriggers the workflow. Proper fix (per-runner
+# DOCKER_CONFIG owned by 1001, internal#597 --env HOME=/home/runner pattern)
+# is tracked as a CI-hygiene follow-up — not in scope here.

 permissions:
  contents: read
  packages: write

 env:
-  IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform
-  TENANT_IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform-tenant
+  # SSOT-Instance-10 (#333): ECR registry triplet (account.dkr.ecr.region.amazonaws.com)
+  # sourced from org/repo var `ECR_REGISTRY` with the current prod-account literal as
+  # bootstrap fallback. When the org var is set, the fallback becomes dead code and
+  # switching accounts/regions is a one-line change at the org level (instead of
+  # touching every workflow). Pattern mirrors `vars.CP_URL || 'literal'` already in
+  # use below in this repo's staging-verify.yml.
+  IMAGE_NAME: ${{ vars.ECR_REGISTRY || '153263036946.dkr.ecr.us-east-2.amazonaws.com' }}/molecule-ai/platform
+  TENANT_IMAGE_NAME: ${{ vars.ECR_REGISTRY || '153263036946.dkr.ecr.us-east-2.amazonaws.com' }}/molecule-ai/platform-tenant

 jobs:
  build-and-push:
-    runs-on: ubuntu-latest
+    # Dedicated publish/release lane (internal#462 / #394 / #399). This
+    # is a post-merge ship job (on: push:main) — it must NOT FIFO-compete
+    # with PR required-CI on the shared pool (PR#1350's prod image build
+    # was delayed ~25min this way). The `publish` label resolves ONLY to
+    # the reserved molecule-runner-publish-* sub-pool (config.publish.yaml,
+    # OUTSIDE the managed 1..20 range) so a merged fix's image build
+    # starts immediately while PR-CI keeps the general pool.
+    runs-on: publish
    steps:
      - name: Checkout
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
@@ -181,7 +202,9 @@ jobs:
    name: Production auto-deploy
    needs: build-and-push
    if: ${{ github.event_name == 'push' && github.ref == 'refs/heads/main' }}
-    runs-on: ubuntu-latest
+    # Publish/release lane (internal#462) — production deploy of a merged
+    # fix; reserved capacity, never queued behind PR-CI.
+    runs-on: publish
    timeout-minutes: 75
    env:
      CP_URL: ${{ vars.PROD_CP_URL || 'https://api.moleculesai.app' }}
@@ -89,6 +89,7 @@ on:
 permissions:
  contents: read
  pull-requests: read
+  secrets: read

 jobs:
  # bp-exempt: PR review bot signal; required merge state is enforced by CI / all-required.
@@ -68,7 +68,10 @@ jobs:
  # bp-exempt: production redeploy is a side-effect workflow, not a merge gate.
  redeploy:
    if: ${{ github.event_name == 'workflow_dispatch' }}
-    runs-on: ubuntu-latest
+    # Dedicated publish/release lane (internal#462 / #394 / #399).
+    # Production tenant redeploy — a deploy action, reserved capacity so
+    # it never queues behind PR-CI. `publish` -> molecule-runner-publish-*.
+    runs-on: publish
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -148,6 +151,11 @@ jobs:
            exit 1
          fi

+          # confirm:true ack required by CP /cp/admin/tenants/redeploy-fleet
+          # contract (cp#228 / task #308) for fleet-wide intent. Empty body
+          # / {confirm:false} / {only_slugs:[]} → 400. This caller redeploys
+          # the entire prod fleet (canary + fan-out), no slug scoping, so
+          # confirm:true is correct.
          BODY=$(jq -nc \
            --arg tag "$TARGET_TAG" \
            --arg canary "$CANARY_SLUG" \
@@ -159,7 +167,8 @@ jobs:
              canary_slug: $canary,
              soak_seconds: $soak,
              batch_size: $batch,
-              dry_run: $dry
+              dry_run: $dry,
+              confirm: true
            }')

          echo "POST $CP_URL/cp/admin/tenants/redeploy-fleet"
@@ -75,7 +75,10 @@ env:
 jobs:
  # bp-exempt: post-merge staging redeploy side effect; CI / all-required gates source changes.
  redeploy:
-    runs-on: ubuntu-latest
+    # Dedicated publish/release lane (internal#462 / #394 / #399).
+    # Post-merge staging redeploy — a deploy action, reserved capacity.
+    # `publish` -> molecule-runner-publish-* sub-pool.
+    runs-on: publish
    # Phase 3 (RFC #219 §1): surface broken workflows without blocking.
    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
    continue-on-error: true
@@ -120,6 +123,11 @@ jobs:
            exit 1
          fi

+          # confirm:true ack required by CP /cp/admin/tenants/redeploy-fleet
+          # contract (cp#228 / task #308) for fleet-wide intent. Empty body
+          # / {confirm:false} / {only_slugs:[]} → 400. Staging IS the
+          # canary, no slug scoping; this rolls the entire staging fleet,
+          # so confirm:true is correct.
          BODY=$(jq -nc \
            --arg tag "$TARGET_TAG" \
            --arg canary "$CANARY_SLUG" \
@@ -131,7 +139,8 @@ jobs:
              canary_slug: $canary,
              soak_seconds: $soak,
              batch_size: $batch,
-              dry_run: $dry
+              dry_run: $dry,
+              confirm: true
            }')

          echo "POST $CP_URL/cp/admin/tenants/redeploy-fleet"
@@ -1,11 +1,16 @@
-# Consolidated comment dispatcher for manual review/tier refires.
+# DEPRECATED — superseded by `.gitea/workflows/sop-checklist.yml`.
 #
+# The review-refire logic (qa/security/tier slash-command dispatch) has been
+# merged into sop-checklist.yml as the `review-refire` job. This workflow
+# is kept as a no-op stub to avoid a gap during the transition window where
+# this file may be deleted while sop-checklist.yml has not yet been merged.
+#
+# After sop-checklist.yml lands, this file will be deleted (issue #1280).
+#
+# Historical behavior (superseded):
 # Gitea 1.22 queues one run per workflow subscribed to `issue_comment` before
-# evaluating job-level `if:`. SOP-heavy PRs therefore created queue storms when
-# qa-review, security-review, sop-checklist, and sop-tier-refire all
-# listened to comments. This workflow is the single non-SOP comment subscriber:
-# ordinary comments no-op quickly; slash commands post the required status
-# contexts to the PR head SHA.
+# evaluating job-level `if:`. Previously this workflow was the single
+# non-SOP comment subscriber for qa/security/tier refire slash commands.

 name: review-refire-comments

@@ -23,91 +28,12 @@ concurrency:
  cancel-in-progress: true

 jobs:
+  # No-op stub — all refire logic moved to sop-checklist.yml review-refire job.
+  # Kept to avoid transition gap; will be deleted after sop-checklist.yml merges.
  dispatch:
    runs-on: ubuntu-latest
    steps:
-      - name: Classify comment
-        id: classify
-        env:
-          COMMENT_BODY: ${{ github.event.comment.body }}
-          IS_PR: ${{ github.event.issue.pull_request != null }}
+      - name: Deprecated — refire logic moved to sop-checklist.yml
        run: |
-          set -euo pipefail
-          {
-            echo "run_qa=false"
-            echo "run_security=false"
-            echo "run_tier=false"
-          } >> "$GITHUB_OUTPUT"
-          if [ "$IS_PR" != "true" ]; then
-            echo "::notice::not a PR comment; no-op"
-            exit 0
-          fi
-          first_line=$(printf '%s\n' "$COMMENT_BODY" | sed -n '1p')
-          case "$first_line" in
-            /qa-recheck*)
-              echo "run_qa=true" >> "$GITHUB_OUTPUT"
-              ;;
-            /security-recheck*)
-              echo "run_security=true" >> "$GITHUB_OUTPUT"
-              ;;
-            /refire-tier-check*)
-              echo "run_tier=true" >> "$GITHUB_OUTPUT"
-              ;;
-            *)
-              echo "::notice::no supported review refire slash command; no-op"
-              ;;
-          esac
-
-      - name: Check out BASE ref for trusted scripts
-        if: |
-          steps.classify.outputs.run_qa == 'true' ||
-          steps.classify.outputs.run_security == 'true' ||
-          steps.classify.outputs.run_tier == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
-        with:
-          ref: ${{ github.event.repository.default_branch }}
-
-      - name: Refire qa-review status
-        if: steps.classify.outputs.run_qa == 'true'
-        env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
-          GITEA_HOST: git.moleculesai.app
-          REPO: ${{ github.repository }}
-          PR_NUMBER: ${{ github.event.issue.number }}
-          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
-          TEAM: qa
-          TEAM_ID: '20'
-          REVIEW_CHECK_DEBUG: '0'
-          REVIEW_CHECK_STRICT: '0'
-          COMMENT_AUTHOR: ${{ github.event.comment.user.login }}
-        run: |
-          set -euo pipefail
-          .gitea/scripts/review-refire-status.sh
-
-      - name: Refire security-review status
-        if: steps.classify.outputs.run_security == 'true'
-        env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
-          GITEA_HOST: git.moleculesai.app
-          REPO: ${{ github.repository }}
-          PR_NUMBER: ${{ github.event.issue.number }}
-          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
-          TEAM: security
-          TEAM_ID: '21'
-          REVIEW_CHECK_DEBUG: '0'
-          REVIEW_CHECK_STRICT: '0'
-          COMMENT_AUTHOR: ${{ github.event.comment.user.login }}
-        run: |
-          set -euo pipefail
-          .gitea/scripts/review-refire-status.sh
-
-      - name: Refire sop-tier-check status
-        if: steps.classify.outputs.run_tier == 'true'
-        env:
-          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
-          GITEA_HOST: git.moleculesai.app
-          REPO: ${{ github.repository }}
-          PR_NUMBER: ${{ github.event.issue.number }}
-          COMMENT_AUTHOR: ${{ github.event.comment.user.login }}
-          SOP_DEBUG: '0'
-        run: bash .gitea/scripts/sop-tier-refire.sh
+          echo "::warning::review-refire-comments.yml is deprecated. Refire logic is now in sop-checklist.yml review-refire job. This workflow is a no-op stub pending deletion (issue #1280)."
+          exit 0
@@ -30,6 +30,11 @@ jobs:
  scan:
    name: Scan diff for credential-shaped strings
    runs-on: ubuntu-latest
+    # Hard CI gate — must complete or the PR is unmergable. 10-minute ceiling
+    # is generous for a diff-scan against a single SHA. If this times out, the
+    # runner is frozen and holding a slot — the step timeout triggers clean
+    # failure, releasing the runner for the next job.
+    timeout-minutes: 10
    steps:
      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
@@ -133,6 +138,14 @@ jobs:
            [ -z "$f" ] && continue
            [ "$f" = "$SELF_GITHUB" ] && continue
            [ "$f" = "$SELF_GITEA" ] && continue
+            # Test-fixture exclude (internal#425): the secrets-detector's OWN
+            # unit-test corpus deliberately embeds credential-SHAPED example
+            # strings to exercise the detector. Verified 2026-05-18 synthetic
+            # (fabricated ghp_* fixtures, not real). Without this the scanner
+            # self-trips on its own fixtures and fail-closes every deploy.
+            # Same rationale as the SELF_* excludes above; gate NOT weakened
+            # (all other paths still fully scanned).
+            [ "$f" = "workspace-server/internal/secrets/patterns_test.go" ] && continue
            if [ -n "$DIFF_RANGE" ]; then
              ADDED=$(git diff --no-color --unified=0 "$BASE" "$HEAD" -- "$f" 2>/dev/null | grep -E '^\+[^+]' || true)
            else
@@ -16,6 +16,7 @@ on:
 permissions:
  contents: read
  pull-requests: read
+  secrets: read

 jobs:
  # bp-exempt: PR security review bot signal; required merge state is enforced by CI / all-required.
@@ -2,24 +2,20 @@
 #
 # RFC#351 Step 2 of 6 (implementation MVP).
 #
-# === DESIGN ===
+# === CONSOLIDATION (issue #1280) ===
 #
-# Goal: each PR must answer 7 SOP-checklist questions in its body,
-# and each item must have at least one /sop-ack <slug> comment from
-# a non-author peer in the required team. BP requires the
-# `sop-checklist / all-items-acked (pull_request)` status to merge.
+# This workflow is the SINGLE `issue_comment` subscriber — the logic from
+# `review-refire-comments.yml` has been merged in. Before this change:
+#   - sop-checklist.yml (pre-2026-05-16) → issue_comment:[created,edited,deleted] → runner slot used, job no-oped
+#   - review-refire-comments.yml           → issue_comment:[created]            → runner slot used, job no-oped
+#   → every non-refire comment occupied 2 runner slots for ~800 s each
+#      (~650 no-op runs/day, ~1,300 runner-slot-occupancy-hours/day).
 #
-# Triggers:
-#   - `pull_request_target`: opened, edited, synchronize, reopened
-#       → fires when PR opens, body is edited (refire — RFC#351 §4),
-#         or new code is pushed (head.sha changes → stale status would
-#         be auto-discarded by BP via dismiss_stale_reviews, but the
-#         status itself is per-SHA so we re-post on the new head).
-#   - `issue_comment`: created, edited, deleted
-#       → fires on any new comment so /sop-ack / /sop-revoke take
-#         effect immediately (Gitea 1.22.6 doesn't refire on
-#         pull_request_review per feedback_pull_request_review_no_refire,
-#         so issue_comment is the canonical refire channel).
+# Fix (PR #1345 / issue #1280):
+#   - ONE workflow, ONE issue_comment:[created] subscription (no edited/deleted)
+#   - all-items-acked job: pull_request_target OR sop slash-command comments
+#   - review-refire job: qa/security/tier refire slash commands
+#   → ~50% reduction in comment-triggered runner occupancy vs pre-fix.
 #
 # Trust boundary (mirrors RFC#324 §A4 + sop-tier-check security note):
 #   `pull_request_target` (not `pull_request`) — workflow def is loaded
@@ -51,7 +47,7 @@
 #   /sop-ack <slug-or-numeric-alias> [optional note]
 #       — register a peer-ack for one checklist item.
 #       — slug accepts kebab-case, snake_case, or natural-spaces
-#         (all normalize to canonical kebab-case).
+#         (all normalized to canonical kebab-case).
 #       — numeric 1..7 maps via config.items[*].numeric_alias.
 #       — most-recent (user, slug) directive wins.
 #
@@ -61,6 +57,13 @@
 #       — most-recent (user, slug) directive wins, so a later /sop-ack
 #         re-restores the ack.
 #
+#   /sop-n/a <gate> [reason]
+#       — declare a gate (qa-review, security-review) N/A.
+#       — see sop-checklist-config.yaml n/a_gates section.
+#
+#   /qa-recheck /security-recheck /refire-tier-check
+#       — refire the corresponding status check on the PR head.
+#
 # The eval is read-only + idempotent (read PR + comments + team
 # membership, compute, post status). Re-running on any event is safe —
 # the new status overwrites the previous one for the same context.
@@ -79,22 +82,22 @@ on:
  pull_request_target:
    types: [opened, edited, synchronize, reopened, labeled, unlabeled]
  issue_comment:
-    types: [created, edited, deleted]
+    types: [created]   # NOT [created, edited, deleted] — Gitea 1.22.6 holds a runner slot
+    # at job-parsing time, before job-level if: guards run. edited/deleted events
+    # occupied ~1,300 runner-slot-hours/day on this workflow alone during the
+    # 2026-05-16 freeze. Per PR #1345 fix.

 permissions:
  contents: read
  pull-requests: read
-  # NOTE: `statuses: write` is the GitHub-Actions name for POST /statuses.
-  # Gitea 1.22.6 may not gate on this permission key (it just checks the
-  # token), but listing it explicitly documents intent for the next
-  # platform-version upgrade.
  statuses: write
+  secrets: read

 jobs:
+  # sop-checklist gate: runs on PR lifecycle events OR sop slash commands.
+  # All other comment types (no-op text comments) no longer assign a runner
+  # because this job's if: guard short-circuits before runner assignment.
  all-items-acked:
-    # Run on pull_request_target events always. On issue_comment events,
-    # only when the comment is on a PR (issue_comment fires for issues
-    # too) and the body contains one of the slash-commands.
    if: |
      github.event_name == 'pull_request_target' ||
      (github.event_name == 'issue_comment' &&
@@ -128,3 +131,95 @@ jobs:
            --pr "$PR_NUMBER" \
            --config .gitea/sop-checklist-config.yaml \
            --gitea-host git.moleculesai.app
+
+  # bp-exempt: informational refire handler, not a merge gate. Emits
+  # qa-review/security-review status updates on /qa-recheck et al slash commands.
+  review-refire:
+    if: |
+      github.event_name == 'issue_comment' &&
+      github.event.issue.pull_request != null
+    runs-on: ubuntu-latest
+    steps:
+      - name: Classify comment
+        id: classify
+        env:
+          COMMENT_BODY: ${{ github.event.comment.body }}
+        run: |
+          set -euo pipefail
+          {
+            echo "run_qa=false"
+            echo "run_security=false"
+            echo "run_tier=false"
+          } >> "$GITHUB_OUTPUT"
+          first_line=$(printf '%s\n' "$COMMENT_BODY" | sed -n '1p')
+          case "$first_line" in
+            /qa-recheck*)
+              echo "run_qa=true" >> "$GITHUB_OUTPUT"
+              ;;
+            /security-recheck*)
+              echo "run_security=true" >> "$GITHUB_OUTPUT"
+              ;;
+            /refire-tier-check*)
+              echo "run_tier=true" >> "$GITHUB_OUTPUT"
+              ;;
+            *)
+              echo "::notice::no supported review refire slash command; no-op"
+              ;;
+          esac
+
+      - name: Check out BASE ref for trusted scripts
+        if: |
+          steps.classify.outputs.run_qa == 'true' ||
+          steps.classify.outputs.run_security == 'true' ||
+          steps.classify.outputs.run_tier == 'true'
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
+        with:
+          ref: ${{ github.event.repository.default_branch }}
+
+      - name: Refire qa-review status
+        if: steps.classify.outputs.run_qa == 'true'
+        env:
+          # RFC_324_TEAM_READ_TOKEN is read-only (team membership read scope only).
+          # review-refire-status.sh POSTs to /statuses — requires write scope.
+          # SOP_TIER_CHECK_TOKEN carries write:repository + write:issue + read:organization.
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_HOST: git.moleculesai.app
+          REPO: ${{ github.repository }}
+          PR_NUMBER: ${{ github.event.issue.number }}
+          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
+          TEAM: qa
+          TEAM_ID: '20'
+          REVIEW_CHECK_DEBUG: '0'
+          REVIEW_CHECK_STRICT: '0'
+        run: |
+          set -euo pipefail
+          .gitea/scripts/review-refire-status.sh
+
+      - name: Refire security-review status
+        if: steps.classify.outputs.run_security == 'true'
+        env:
+          # RFC_324_TEAM_READ_TOKEN is read-only (team membership read scope only).
+          # review-refire-status.sh POSTs to /statuses — requires write scope.
+          # SOP_TIER_CHECK_TOKEN carries write:repository + write:issue + read:organization.
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_HOST: git.moleculesai.app
+          REPO: ${{ github.repository }}
+          PR_NUMBER: ${{ github.event.issue.number }}
+          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
+          TEAM: security
+          TEAM_ID: '21'
+          REVIEW_CHECK_DEBUG: '0'
+          REVIEW_CHECK_STRICT: '0'
+        run: |
+          set -euo pipefail
+          .gitea/scripts/review-refire-status.sh
+
+      - name: Refire sop-tier-check status
+        if: steps.classify.outputs.run_tier == 'true'
+        env:
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_HOST: git.moleculesai.app
+          REPO: ${{ github.repository }}
+          PR_NUMBER: ${{ github.event.issue.number }}
+          SOP_DEBUG: '0'
+        run: bash .gitea/scripts/sop-tier-refire.sh
@@ -71,6 +71,7 @@ jobs:
    permissions:
      contents: read
      pull-requests: read
+      secrets: read
    steps:
      - name: Check out base branch (for the script)
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
@@ -75,8 +75,12 @@ permissions:
 env:
  # ECR registry (post-2026-05-06 SSOT for tenant images).
  # publish-workspace-server-image.yml pushes here.
-  IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform
-  TENANT_IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform-tenant
+  # SSOT-Instance-10 (#333): triplet sourced from org/repo var `ECR_REGISTRY` with
+  # the current prod-account literal as bootstrap fallback. When the org var is set,
+  # the fallback becomes dead code and switching accounts/regions is a one-line
+  # change at the org level. Pattern mirrors `vars.CP_URL || 'literal'` below.
+  IMAGE_NAME: ${{ vars.ECR_REGISTRY || '153263036946.dkr.ecr.us-east-2.amazonaws.com' }}/molecule-ai/platform
+  TENANT_IMAGE_NAME: ${{ vars.ECR_REGISTRY || '153263036946.dkr.ecr.us-east-2.amazonaws.com' }}/molecule-ai/platform-tenant
  # CP endpoint for redeploy-fleet (used in promote step below).
  CP_URL: ${{ vars.CP_URL || 'https://staging-api.moleculesai.app' }}
  GITHUB_SERVER_URL: https://git.moleculesai.app
@@ -235,6 +239,11 @@ jobs:
          set -euo pipefail

          TARGET_TAG="staging-${SHA}"
+          # confirm:true ack required by CP /cp/admin/tenants/redeploy-fleet
+          # contract (cp#228 / task #308) for fleet-wide intent. Empty body
+          # / {confirm:false} / {only_slugs:[]} → 400. This caller promotes
+          # the verified staging image across the entire prod fleet (canary
+          # + fan-out), no slug scoping, so confirm:true is correct.
          BODY=$(jq -nc \
            --arg tag "$TARGET_TAG" \
            --argjson soak "${SOAK_SECONDS:-120}" \
@@ -244,7 +253,8 @@ jobs:
              target_tag: $tag,
              soak_seconds: $soak,
              batch_size: $batch,
-              dry_run: $dry
+              dry_run: $dry,
+              confirm: true
            }')

          if [ -n "${CANARY_SLUG:-}" ]; then
@@ -53,19 +53,12 @@ name: status-reaper
 # `inputs:` block here. Gitea 1.22.6 rejects the whole workflow as
 # "unknown on type" when `workflow_dispatch.inputs.X` is present.
 on:
-  # SCHEDULE RE-ENABLED 2026-05-12 rev3 — interim disable (mc#645) reverted now that
-  # rev3 widens DEFAULT_SWEEP_LIMIT 10 → 30 (covers retroactive-failure timing window).
-  # Sibling watchdog re-enabled in the same PR with timeout-minutes raised 5 → 15.
-  schedule:
-    # Every 5 minutes. Off-zero alignment with sibling cron workflows:
-    # ci-required-drift (`:17`), main-red-watchdog (`:05`),
-    # railway-pin-audit (`:23`). 5-min cadence gives a tight enough
-    # close on schedule-triggered false-reds that main-red-watchdog
-    # (hourly :05) almost never files an issue on the false case.
-    # rev3 keeps `*/5` unchanged per hongming-pc2 03:25Z review:
-    # "trades window-width-cheap for cadence-loady" — N=30 widens
-    # the lookback cheaply without doubling runner load via `*/2`.
-    - cron: '*/5 * * * *'
+  # Schedule moved to operator-config:
+  #   /etc/cron.d/molecule-core-status-reaper ->
+  #   /usr/local/bin/molecule-core-cron-bot.sh status-reaper
+  #
+  # This keeps the 5-minute compensation cadence but stops a maintenance
+  # bot from consuming Gitea Actions runner slots during PR merge waves.
  workflow_dispatch:

 # Compensating-status POST needs write on repo statuses; no other
@@ -1,154 +0,0 @@
-name: Block internal-flavored paths
-
-# Hard CI gate. Internal content (positioning, competitive briefs, sales
-# playbooks, PMM/press drip, draft campaigns) lives in molecule-ai/internal —
-# this public monorepo must never re-acquire those paths. CEO directive
-# 2026-04-23 after a fleet-wide audit found 79 internal files leaked here.
-#
-# Failure mode without this gate: agents (PMM, Research, DevRel, Sales) drop
-# briefs into the easiest path their cwd resolves to (root /research,
-# /marketing, /docs/marketing) and gitignore alone won't catch a `git add -f`
-# or a stale gitignore line. This workflow is the mechanical backstop.
-
-on:
-  pull_request:
-    types: [opened, synchronize, reopened]
-  push:
-    branches: [main, staging]
-  # Required for GitHub merge queue: the queue's pre-merge CI run on
-  # `gh-readonly-queue/...` refs needs this check to fire so the queue
-  # gets a real result instead of stalling forever AWAITING_CHECKS.
-  merge_group:
-    types: [checks_requested]
-
-jobs:
-  check:
-    name: Block forbidden paths
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-        with:
-          fetch-depth: 2  # need previous commit to diff against on push events
-
-      # For pull_request events the diff base is github.event.pull_request.base.sha,
-      # which may be many commits behind HEAD and therefore absent from the
-      # shallow clone above.  Fetch it explicitly (depth=1 keeps it fast).
-      - name: Fetch PR base SHA (pull_request events only)
-        if: github.event_name == 'pull_request'
-        run: git fetch --depth=1 origin ${{ github.event.pull_request.base.sha }}
-
-      # For merge_group events the queue's pre-merge ref is a commit on
-      # `gh-readonly-queue/...` whose parent is the queue's base_sha.
-      # That parent isn't part of the queue branch's shallow clone, so
-      # we fetch it explicitly. Mirrors the equivalent step in
-      # secret-scan.yml (#2120) — same shallow-clone bug class.
-      - name: Fetch merge_group base SHA (merge_group events only)
-        if: github.event_name == 'merge_group'
-        run: git fetch --depth=1 origin ${{ github.event.merge_group.base_sha }}
-
-      - name: Refuse if forbidden paths appear
-        env:
-          # Plumb event-specific SHAs through env so the script doesn't
-          # need conditional `${{ ... }}` interpolation per event type.
-          # github.event.before/after only exist on push events;
-          # merge_group has its own base_sha/head_sha; pull_request has
-          # pull_request.base.sha / pull_request.head.sha.
-          PR_BASE_SHA: ${{ github.event.pull_request.base.sha }}
-          PR_HEAD_SHA: ${{ github.event.pull_request.head.sha }}
-          MG_BASE_SHA: ${{ github.event.merge_group.base_sha }}
-          MG_HEAD_SHA: ${{ github.event.merge_group.head_sha }}
-          PUSH_BEFORE: ${{ github.event.before }}
-          PUSH_AFTER: ${{ github.event.after }}
-        run: |
-          # Paths that must NEVER live in the public monorepo. Add to this
-          # list narrowly — broader patterns belong in .gitignore so day-to-day
-          # docs work isn't accidentally blocked.
-          FORBIDDEN_PATTERNS=(
-            "^research/"
-            "^marketing/"
-            "^docs/marketing/"
-            "^comment-[0-9]+\.json$"
-            "^test-pmm.*\.(txt|md)$"
-            "^tick-reflections.*\.(txt|md)$"
-            ".*-temp\.(md|txt)$"
-          )
-
-          # Determine the diff base. Each event type stores its SHAs in
-          # a different place — see the env block above.
-          case "${{ github.event_name }}" in
-            pull_request)
-              BASE="$PR_BASE_SHA"
-              HEAD="$PR_HEAD_SHA"
-              ;;
-            merge_group)
-              BASE="$MG_BASE_SHA"
-              HEAD="$MG_HEAD_SHA"
-              ;;
-            *)
-              BASE="$PUSH_BEFORE"
-              HEAD="$PUSH_AFTER"
-              ;;
-          esac
-
-          # On push events with shallow clones, BASE may be present in
-          # the event payload but absent from the local object DB
-          # (fetch-depth=2 doesn't always reach the previous commit
-          # across true merges). Try fetching it on demand. If the
-          # fetch fails — e.g. the SHA was force-overwritten — we fall
-          # through to the empty-BASE branch below, which scans the
-          # entire tree as if every file were new. Correct, just slow.
-          # Same recovery shape as secret-scan.yml (#2120 — incident
-          # 2026-04-27 06:50Z block-internal-paths exit 128 with
-          # "fatal: bad object <sha>" on staging push).
-          if [ -n "$BASE" ] && ! echo "$BASE" | grep -qE '^0+$'; then
-            if ! git cat-file -e "$BASE" 2>/dev/null; then
-              git fetch --depth=1 origin "$BASE" 2>/dev/null || true
-            fi
-          fi
-
-          # Files added or modified in this change.
-          if [ -z "$BASE" ] || echo "$BASE" | grep -qE '^0+$' || ! git cat-file -e "$BASE" 2>/dev/null; then
-            # New branch / no previous SHA / BASE unreachable — check
-            # the entire tree as if every file were new. Slower but
-            # correct on first push or post-fetch-failure recovery.
-            CHANGED=$(git ls-tree -r --name-only HEAD)
-          else
-            CHANGED=$(git diff --name-only --diff-filter=AM "$BASE" "$HEAD")
-          fi
-
-          if [ -z "$CHANGED" ]; then
-            echo "No changed files to inspect."
-            exit 0
-          fi
-
-          OFFENDING=""
-          for path in $CHANGED; do
-            for pattern in "${FORBIDDEN_PATTERNS[@]}"; do
-              if echo "$path" | grep -qE "$pattern"; then
-                OFFENDING="${OFFENDING}${path} (matched: ${pattern})\n"
-                break
-              fi
-            done
-          done
-
-          if [ -n "$OFFENDING" ]; then
-            echo "::error::Forbidden internal-flavored paths detected:"
-            printf "$OFFENDING"
-            echo ""
-            echo "These paths belong in molecule-ai/internal, not this public repo."
-            echo "See docs/internal-content-policy.md for canonical locations."
-            echo ""
-            echo "If your file is genuinely public-facing (e.g. a blog post"
-            echo "ready to ship), use one of these alternatives instead:"
-            echo "  • Public-bound blog posts:  docs/blog/<slug>.md"
-            echo "  • Public-bound tutorials:   docs/tutorials/<slug>.md"
-            echo "  • Public devrel content:    docs/devrel/<slug>.md"
-            echo ""
-            echo "If you legitimately need to add a new top-level path that"
-            echo "happens to match a forbidden pattern, edit"
-            echo ".github/workflows/block-internal-paths.yml and update the"
-            echo "FORBIDDEN_PATTERNS list with reviewer signoff."
-            exit 1
-          fi
-
-          echo "✓ No forbidden paths in this change."
@@ -1,320 +0,0 @@
-name: Canary — staging SaaS smoke (every 30 min)
-
-# Minimum viable health check: provisions one Hermes workspace on a fresh
-# staging org, sends one A2A message, verifies PONG, tears down. ~8 min
-# wall clock. Pages on failure by opening a GitHub issue; auto-closes the
-# issue on the next green run.
-#
-# The full-SaaS workflow (e2e-staging-saas.yml) covers the broader surface
-# but runs only on provisioning-critical pushes + nightly — this one
-# catches drift in the 30-min window between those runs (AMI health, CF
-# cert rotation, WorkOS session stability, etc.).
-#
-# Lean mode: E2E_MODE=canary skips the child workspace + HMA memory +
-# peers/activity checks. One parent workspace + one A2A turn is enough
-# to signal "SaaS stack end-to-end is alive."
-
-on:
-  schedule:
-    # Every 30 min. Cron on GitHub-hosted runners has a known drift of
-    # a few minutes under load — that's fine for a canary.
-    - cron: '*/30 * * * *'
-  workflow_dispatch:
-    inputs:
-      keep_on_failure:
-        description: >-
-          Skip teardown when the canary fails (debugging only). The
-          tenant org + EC2 + CF tunnel + DNS stay alive so an operator
-          can SSM into the workspace EC2 and capture docker logs of the
-          failing claude-code container. REMEMBER to manually delete
-          via DELETE /cp/admin/tenants/<slug> when done so the org
-          doesn't accumulate cost. Only honored on workflow_dispatch;
-          cron runs always tear down (we don't want unattended cron
-          to leak resources).
-        type: boolean
-        default: false
-
-# Serialise with the full-SaaS workflow so they don't contend for the
-# same org-create quota on staging. Different group key from
-# e2e-staging-saas since we don't mind queueing canaries behind one
-# full run, but two canaries SHOULD queue against each other.
-concurrency:
-  group: canary-staging
-  cancel-in-progress: false
-
-permissions:
-  # Needed to open / close the alerting issue.
-  issues: write
-  contents: read
-
-jobs:
-  canary:
-    name: Canary smoke
-    runs-on: ubuntu-latest
-    # 25 min headroom over the 15-min TLS-readiness deadline in
-    # tests/e2e/test_staging_full_saas.sh (#2107). Without the buffer
-    # the job is killed at the wall-clock 15:00 mark BEFORE the bash
-    # `fail` + diagnostic burst can fire, leaving every cancellation
-    # silent. Sibling staging E2E jobs run at 20-45 min — keeping
-    # canary tighter than them so a true wedge still surfaces here
-    # first.
-    timeout-minutes: 25
-
-    env:
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-      # MiniMax is the canary's PRIMARY LLM auth path post-2026-05-04.
-      # Switched from hermes+OpenAI after #2578 (the staging OpenAI key
-      # account went over quota and stayed dead for 36+ hours, taking
-      # the canary red the entire time). claude-code template's
-      # `minimax` provider routes ANTHROPIC_BASE_URL to
-      # api.minimax.io/anthropic and reads MINIMAX_API_KEY at boot —
-      # ~5-10x cheaper per token than gpt-4.1-mini AND on a separate
-      # billing account, so OpenAI quota collapse no longer wedges the
-      # canary. Mirrors the migration continuous-synth-e2e.yml made on
-      # 2026-05-03 (#265) for the same reason. tests/e2e/test_staging_
-      # full_saas.sh branches SECRETS_JSON on which key is present —
-      # MiniMax wins when set.
-      E2E_MINIMAX_API_KEY: ${{ secrets.MOLECULE_STAGING_MINIMAX_API_KEY }}
-      # Direct-Anthropic alternative for operators who don't want to
-      # set up a MiniMax account (priority below MiniMax — first
-      # non-empty wins in test_staging_full_saas.sh's secrets-injection
-      # block). See #2578 PR comment for the rationale.
-      E2E_ANTHROPIC_API_KEY: ${{ secrets.MOLECULE_STAGING_ANTHROPIC_API_KEY }}
-      # OpenAI fallback — kept wired so an operator-dispatched run with
-      # E2E_RUNTIME=hermes overridden via workflow_dispatch can still
-      # exercise the OpenAI path without re-editing the workflow.
-      E2E_OPENAI_API_KEY: ${{ secrets.MOLECULE_STAGING_OPENAI_KEY }}
-      E2E_MODE: canary
-      E2E_RUNTIME: claude-code
-      # Pin the canary to a specific MiniMax model rather than relying
-      # on the per-runtime default (which could resolve to "sonnet" →
-      # direct Anthropic and defeat the cost saving). M2.7-highspeed
-      # is "Token Plan only" but cheap-per-token and fast.
-      E2E_MODEL_SLUG: MiniMax-M2.7-highspeed
-      E2E_RUN_ID: "canary-${{ github.run_id }}"
-      # Debug-only: when an operator dispatches with keep_on_failure=true,
-      # the canary script's E2E_KEEP_ORG=1 path skips teardown so the
-      # tenant org + EC2 stay alive for SSM-based log capture. Cron runs
-      # never set this (the input only exists on workflow_dispatch) so
-      # unattended cron always tears down. See molecule-core#129
-      # failure mode #1 — capturing the actual exception requires
-      # docker logs from the live container.
-      E2E_KEEP_ORG: ${{ github.event.inputs.keep_on_failure == 'true' && '1' || '0' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify admin token present
-        run: |
-          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
-            echo "::error::MOLECULE_STAGING_ADMIN_TOKEN not set"
-            exit 2
-          fi
-
-      - name: Verify LLM key present
-        run: |
-          # Per-runtime key check — claude-code uses MiniMax; hermes /
-          # langgraph (operator-dispatched only) use OpenAI. Hard-fail
-          # rather than soft-skip per the lesson from synth E2E #2578:
-          # an empty key silently falls through to the wrong
-          # SECRETS_JSON branch and the canary fails 5 min later with
-          # a confusing auth error instead of the clean "secret
-          # missing" message at the top.
-          case "${E2E_RUNTIME}" in
-            claude-code)
-              # Either MiniMax OR direct-Anthropic works — first
-              # non-empty wins in the test script's secrets-injection
-              # priority chain. Operators only need to set ONE of these
-              # secrets; we don't force a choice between them.
-              if [ -n "${E2E_MINIMAX_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY"
-                required_secret_value="${E2E_MINIMAX_API_KEY}"
-              elif [ -n "${E2E_ANTHROPIC_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value="${E2E_ANTHROPIC_API_KEY}"
-              else
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY or MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value=""
-              fi
-              ;;
-            langgraph|hermes)
-              required_secret_name="MOLECULE_STAGING_OPENAI_KEY"
-              required_secret_value="${E2E_OPENAI_API_KEY:-}"
-              ;;
-            *)
-              echo "::warning::Unknown E2E_RUNTIME='${E2E_RUNTIME}' — skipping LLM-key check"
-              required_secret_name=""
-              required_secret_value="present"
-              ;;
-          esac
-          if [ -n "$required_secret_name" ] && [ -z "$required_secret_value" ]; then
-            echo "::error::${required_secret_name} secret not set for runtime=${E2E_RUNTIME} — A2A will fail at request time with 'No LLM provider configured'"
-            exit 2
-          fi
-          echo "LLM key present ✓ (runtime=${E2E_RUNTIME}, key=${required_secret_name}, len=${#required_secret_value})"
-
-      - name: Canary run
-        id: canary
-        run: bash tests/e2e/test_staging_full_saas.sh
-
-      # Alerting: open a sticky issue on the FIRST failure; comment on
-      # subsequent failures; auto-close on next green. Comment-on-existing
-      # de-duplicates so a single open issue accumulates the streak —
-      # ops sees one issue with N comments rather than N issues.
-      #
-      # Why no consecutive-failures threshold (e.g., wait 3 runs before
-      # filing): the prior threshold check used
-      # `github.rest.actions.listWorkflowRuns()` which Gitea 1.22.6 does
-      # not expose (returns 404). On Gitea Actions the threshold call
-      # ALWAYS failed, breaking the entire alerting step and going days
-      # silent on real regressions (38h+ chronic red on 2026-05-07/08
-      # before this fix; tracked in molecule-core#129). Filing on first
-      # failure is also better UX — we want to know about the first red,
-      # not wait 90 min for it to "count." Real flakes get one issue +
-      # a quick close-on-green; persistent reds accumulate comments.
-      - name: Open issue on failure
-        if: failure()
-        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
-        with:
-          script: |
-            const title = '🔴 Canary failing: staging SaaS smoke';
-            const runURL = `${context.serverUrl}/${context.repo.owner}/${context.repo.repo}/actions/runs/${context.runId}`;
-
-            // Find an existing open canary issue (stable title match).
-            // If one exists, this isn't a "first failure" — comment and exit.
-            const { data: existing } = await github.rest.issues.listForRepo({
-              owner: context.repo.owner, repo: context.repo.repo,
-              state: 'open', labels: 'canary-staging',
-              per_page: 10,
-            });
-            const match = existing.find(i => i.title === title);
-            if (match) {
-              await github.rest.issues.createComment({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: match.number,
-                body: `Canary still failing. ${runURL}`,
-              });
-              core.info(`Commented on existing issue #${match.number}`);
-              return;
-            }
-
-            // No open issue yet — file one on this first failure. The
-            // comment-on-existing branch above means subsequent failures
-            // accumulate as comments on this same issue, so we don't
-            // spam new issues per run.
-            const body =
-              `Canary run failed at ${new Date().toISOString()}.\n\n` +
-              `Run: ${runURL}\n\n` +
-              `This issue auto-closes on the next green canary run. ` +
-              `Consecutive failures add a comment here rather than a new issue.`;
-            await github.rest.issues.create({
-              owner: context.repo.owner, repo: context.repo.repo,
-              title, body,
-              labels: ['canary-staging', 'bug'],
-            });
-            core.info('Opened canary failure issue (first red)');
-
-      - name: Auto-close canary issue on success
-        if: success()
-        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
-        with:
-          script: |
-            const title = '🔴 Canary failing: staging SaaS smoke';
-            const { data: open } = await github.rest.issues.listForRepo({
-              owner: context.repo.owner, repo: context.repo.repo,
-              state: 'open', labels: 'canary-staging',
-              per_page: 10,
-            });
-            const match = open.find(i => i.title === title);
-            if (match) {
-              await github.rest.issues.createComment({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: match.number,
-                body: `Canary recovered at ${new Date().toISOString()}. Closing.`,
-              });
-              await github.rest.issues.update({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: match.number,
-                state: 'closed',
-              });
-              core.info(`Closed recovered canary issue #${match.number}`);
-            }
-
-      - name: Teardown safety net
-        if: always()
-        env:
-          ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-        run: |
-          set +e
-          # Slug prefix matches what test_staging_full_saas.sh emits
-          # in canary mode:
-          #   SLUG="e2e-canary-$(date +%Y%m%d)-${RUN_ID_SUFFIX}"
-          # Earlier this was `e2e-{today}-canary-` — that was the
-          # full-mode pattern (date FIRST, mode SECOND); canary slugs
-          # have mode FIRST, date SECOND. The mismatch silently
-          # never matched, leaving every cancelled-canary EC2 alive
-          # until the once-an-hour sweep eventually caught it
-          # (incident 2026-04-26 21:03Z: 1h25m EC2 leak before manual
-          # cleanup; same gap on three earlier cancellations today).
-          orgs=$(curl -sS "$MOLECULE_CP_URL/cp/admin/orgs" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" 2>/dev/null \
-            | python3 -c "
-          import json, sys, os, datetime
-          run_id = os.environ.get('GITHUB_RUN_ID', '')
-          d = json.load(sys.stdin)
-          # Scope to slugs from THIS canary run when GITHUB_RUN_ID is
-          # available; the canary workflow sets E2E_RUN_ID='canary-\${run_id}'
-          # so the slug suffix is '-canary-\${run_id}-...'. Mirrors the
-          # full-mode safety net's per-run scoping (e2e-staging-saas.yml)
-          # added after the 2026-04-21 cross-run cleanup incident.
-          # Sweep both today AND yesterday's UTC dates so a run that
-          # crosses midnight still cleans up its own slug — see the
-          # 2026-04-26→27 canvas-safety-net incident.
-          today = datetime.date.today()
-          yesterday = today - datetime.timedelta(days=1)
-          dates = (today.strftime('%Y%m%d'), yesterday.strftime('%Y%m%d'))
-          if run_id:
-              prefixes = tuple(f'e2e-canary-{d}-canary-{run_id}' for d in dates)
-          else:
-              prefixes = tuple(f'e2e-canary-{d}-' for d in dates)
-          candidates = [o['slug'] for o in d.get('orgs', [])
-                        if any(o.get('slug','').startswith(p) for p in prefixes)
-                        and o.get('status') not in ('purged',)]
-          print('\n'.join(candidates))
-          " 2>/dev/null)
-          # Per-slug DELETE with HTTP-code verification. The previous
-          # `... >/dev/null || true` swallowed every failure, so a 5xx
-          # or timeout from CP looked identical to "successfully cleaned
-          # up" and the tenant kept eating ~2 vCPU until the hourly
-          # stale sweep caught it (up to 2h later). Now we capture the
-          # response code and surface non-2xx as a workflow warning, so
-          # the run page shows which slug leaked. We still don't `exit 1`
-          # on cleanup failure — a single-canary cleanup miss shouldn't
-          # fail-flag the canary itself when the actual smoke check
-          # passed. The sweep-stale-e2e-orgs cron (now every 15 min,
-          # 30-min threshold) is the safety net for whatever slips past.
-          # See molecule-controlplane#420.
-          leaks=()
-          for slug in $orgs; do
-            # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-            # pollution of the captured status (lint-curl-status-capture.yml).
-            set +e
-            curl -sS -o /tmp/canary-cleanup.out -w "%{http_code}" \
-              -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-              -H "Authorization: Bearer $ADMIN_TOKEN" \
-              -H "Content-Type: application/json" \
-              -d "{\"confirm\":\"$slug\"}" >/tmp/canary-cleanup.code
-            set -e
-            code=$(cat /tmp/canary-cleanup.code 2>/dev/null || echo "000")
-            if [ "$code" = "200" ] || [ "$code" = "204" ]; then
-              echo "[teardown] deleted $slug (HTTP $code)"
-            else
-              echo "::warning::canary teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within ~45 min. Body: $(head -c 300 /tmp/canary-cleanup.out 2>/dev/null)"
-              leaks+=("$slug")
-            fi
-          done
-          if [ ${#leaks[@]} -gt 0 ]; then
-            echo "::warning::canary teardown left ${#leaks[@]} leak(s): ${leaks[*]}"
-          fi
-          exit 0
@@ -1,255 +0,0 @@
-name: canary-verify
-
-# Runs the canary smoke suite against the staging canary tenant fleet
-# after a new :staging-<sha> image lands in ECR. On green, calls the
-# CP redeploy-fleet endpoint to promote :staging-<sha> → :latest so
-# the prod tenant fleet's 5-minute auto-updater picks up the verified
-# digest. On red, :latest stays on the prior known-good digest and
-# prod is untouched.
-#
-# Registry note (2026-05-10): This workflow previously used GHCR
-# (ghcr.io/molecule-ai/platform-tenant) — that registry was retired
-# during the 2026-05-06 Gitea suspension migration when publish-
-# workspace-server-image.yml switched to the operator's ECR org
-# (153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/
-# platform-tenant). The GHCR → ECR migration was never applied to
-# this file, so canary-verify was silently smoke-testing the stale
-# GHCR image while the actual staging/prod tenants ran the ECR image.
-# Result: smoke tests could not catch a broken ECR build. Fix:
-#   - Wait step: reads SHA from running canary /health (tenant-
-#     agnostic, works regardless of registry).
-#   - Promote step: calls CP redeploy-fleet endpoint with target_tag=
-#     staging-<sha>, same mechanism as redeploy-tenants-on-main.yml.
-#     No longer attempts GHCR crane ops.
-#
-# Dependencies:
-#   - publish-workspace-server-image.yml publishes :staging-<sha>
-#     to ECR on staging and main merges.
-#   - Canary tenants are configured to pull :staging-<sha> from ECR
-#     (TENANT_IMAGE env set to the ECR :staging-<sha> tag).
-#   - Repo secrets CANARY_TENANT_URLS / CANARY_ADMIN_TOKENS /
-#     CANARY_CP_SHARED_SECRET are populated.
-
-on:
-  workflow_run:
-    workflows: ["publish-workspace-server-image"]
-    types: [completed]
-  workflow_dispatch:
-
-permissions:
-  contents: read
-  packages: write
-  actions: read
-
-env:
-  # ECR registry (post-2026-05-06 SSOT for tenant images).
-  # publish-workspace-server-image.yml pushes here.
-  IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform
-  TENANT_IMAGE_NAME: 153263036946.dkr.ecr.us-east-2.amazonaws.com/molecule-ai/platform-tenant
-  # CP endpoint for redeploy-fleet (used in promote step below).
-  CP_URL: ${{ vars.CP_URL || 'https://staging-api.moleculesai.app' }}
-
-jobs:
-  canary-smoke:
-    # Skip when the upstream workflow failed — no image to test against.
-    if: ${{ github.event.workflow_run.conclusion == 'success' || github.event_name == 'workflow_dispatch' }}
-    runs-on: ubuntu-latest
-    outputs:
-      sha: ${{ steps.compute.outputs.sha }}
-      smoke_ran: ${{ steps.smoke.outputs.ran }}
-    steps:
-      - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Compute sha
-        id: compute
-        run: echo "sha=${GITHUB_SHA::7}" >> "$GITHUB_OUTPUT"
-
-      - name: Wait for canary tenants to pick up :staging-<sha>
-        # Poll canary health endpoints every 30s for up to 7 min instead
-        # of a fixed 6-min sleep. Exits as soon as ALL canaries report
-        # the new SHA (~2-3 min typical vs 6 min fixed). Falls back to
-        # proceeding after 7 min even if not all canaries responded —
-        # the smoke suite will catch any that didn't update.
-        #
-        # NOTE: The SHA is read from the running tenant's /health response,
-        # NOT from a registry lookup. This is registry-agnostic and works
-        # regardless of whether the tenant pulls from ECR, GHCR, or any
-        # other registry — the canary is telling us what it's actually
-        # running, which is the ground truth for smoke testing.
-        env:
-          CANARY_TENANT_URLS: ${{ secrets.CANARY_TENANT_URLS }}
-          EXPECTED_SHA: ${{ steps.compute.outputs.sha }}
-        run: |
-          if [ -z "$CANARY_TENANT_URLS" ]; then
-            echo "No canary URLs configured — falling back to 60s wait"
-            sleep 60
-            exit 0
-          fi
-          IFS=',' read -ra URLS <<< "$CANARY_TENANT_URLS"
-          MAX_WAIT=420  # 7 minutes
-          INTERVAL=30
-          ELAPSED=0
-          while [ $ELAPSED -lt $MAX_WAIT ]; do
-            ALL_READY=true
-            for url in "${URLS[@]}"; do
-              HEALTH=$(curl -s --max-time 5 "${url}/health" 2>/dev/null || echo "{}")
-              SHA=$(echo "$HEALTH" | grep -o "\"sha\":\"[^\"]*\"" | head -1 | cut -d'"' -f4)
-              if [ "$SHA" != "$EXPECTED_SHA" ]; then
-                ALL_READY=false
-                break
-              fi
-            done
-            if $ALL_READY; then
-              echo "All canaries running staging-${EXPECTED_SHA} after ${ELAPSED}s"
-              exit 0
-            fi
-            echo "Waiting for canaries... (${ELAPSED}s / ${MAX_WAIT}s)"
-            sleep $INTERVAL
-            ELAPSED=$((ELAPSED + INTERVAL))
-          done
-          echo "Timeout after ${MAX_WAIT}s — proceeding anyway (smoke suite will validate)"
-
-      - name: Run canary smoke suite
-        id: smoke
-        # Graceful-skip when no canary fleet is configured (Phase 2 not yet
-        # stood up — see molecule-controlplane/docs/canary-tenants.md).
-        # Sets `ran=false` on skip so promote-to-latest stays off (we don't
-        # want every main merge auto-promoting without gating). Manual
-        # promote-latest.yml is the release gate while canary is absent.
-        # Once the fleet is real: delete the early-exit branch.
-        env:
-          CANARY_TENANT_URLS: ${{ secrets.CANARY_TENANT_URLS }}
-          CANARY_ADMIN_TOKENS: ${{ secrets.CANARY_ADMIN_TOKENS }}
-          CANARY_CP_BASE_URL: https://staging-api.moleculesai.app
-          CANARY_CP_SHARED_SECRET: ${{ secrets.CANARY_CP_SHARED_SECRET }}
-        run: |
-          set -euo pipefail
-          if [ -z "${CANARY_TENANT_URLS:-}" ] \
-            || [ -z "${CANARY_ADMIN_TOKENS:-}" ] \
-            || [ -z "${CANARY_CP_SHARED_SECRET:-}" ]; then
-            {
-              echo "## ⚠️ canary-verify skipped"
-              echo
-              echo "One or more canary secrets are unset (\`CANARY_TENANT_URLS\`, \`CANARY_ADMIN_TOKENS\`, \`CANARY_CP_SHARED_SECRET\`)."
-              echo "Phase 2 canary fleet has not been stood up yet —"
-              echo "see [canary-tenants.md](https://git.moleculesai.app/molecule-ai/molecule-controlplane/blob/main/docs/canary-tenants.md)."
-              echo
-              echo "**Skipped — promote-to-latest will NOT auto-fire.** Dispatch \`promote-latest.yml\` manually when ready."
-            } >> "$GITHUB_STEP_SUMMARY"
-            echo "ran=false" >> "$GITHUB_OUTPUT"
-            echo "::notice::canary-verify: skipped — no canary fleet configured"
-            exit 0
-          fi
-          bash scripts/canary-smoke.sh
-          echo "ran=true" >> "$GITHUB_OUTPUT"
-
-      - name: Summary on failure
-        if: ${{ failure() }}
-        run: |
-          {
-            echo "## Canary smoke FAILED"
-            echo
-            echo "Canary tenants rejected image \`staging-${{ steps.compute.outputs.sha }}\`."
-            echo ":latest stays pinned to the prior good digest — prod is untouched."
-            echo
-            echo "Fix forward and merge again, or investigate the specific failed"
-            echo "assertions in the canary-smoke step log above."
-          } >> "$GITHUB_STEP_SUMMARY"
-
-  promote-to-latest:
-    # On green, calls the CP redeploy-fleet endpoint with target_tag=
-    # staging-<sha> to promote the verified ECR image. This is the same
-    # mechanism as redeploy-tenants-on-main.yml — no GHCR crane ops.
-    #
-    # Pre-fix history: the old GHCR promote step used `crane tag` against
-    # ghcr.io/molecule-ai/platform-tenant, but publish-workspace-server-
-    # image.yml had already migrated to ECR on 2026-05-07 (commit
-    # 10e510f5). The GHCR tags were never updated, so this step was
-    # silently promoting a stale GHCR image while actual prod tenants
-    # pulled from ECR. Canary smoke tests were GHCR-targeted and could
-    # not catch a broken ECR build.
-    needs: canary-smoke
-    if: ${{ needs.canary-smoke.result == 'success' && needs.canary-smoke.outputs.smoke_ran == 'true' }}
-    runs-on: ubuntu-latest
-    env:
-      SHA: ${{ needs.canary-smoke.outputs.sha }}
-      CP_URL: ${{ vars.CP_URL || 'https://staging-api.moleculesai.app' }}
-      # CP_ADMIN_API_TOKEN gates write access to the redeploy endpoint.
-      # Stored at the repo level so all workflows pick it up automatically.
-      CP_ADMIN_API_TOKEN: ${{ secrets.CP_ADMIN_API_TOKEN }}
-      # canary_slug pin: deploy the verified :staging-<sha> to the canary
-      # first (soak 120s), then fan out to the rest of the fleet.
-      CANARY_SLUG: ${{ vars.CANARY_PROMOTE_SLUG || '' }}
-      SOAK_SECONDS: ${{ vars.CANARY_PROMOTE_SOAK || '120' }}
-      BATCH_SIZE: ${{ vars.CANARY_PROMOTE_BATCH || '3' }}
-    steps:
-      - name: Check CP credentials
-        run: |
-          if [ -z "${CP_ADMIN_API_TOKEN:-}" ]; then
-            echo "::error::CP_ADMIN_API_TOKEN secret is not set — promote step cannot call redeploy-fleet."
-            echo "::error::Set it at: repo Settings → Actions → Variables and Secrets → New Secret."
-            exit 1
-          fi
-
-      - name: Promote verified ECR image to :latest
-        run: |
-          set -euo pipefail
-
-          TARGET_TAG="staging-${SHA}"
-          BODY=$(jq -nc \
-            --arg tag "$TARGET_TAG" \
-            --argjson soak "${SOAK_SECONDS:-120}" \
-            --argjson batch "${BATCH_SIZE:-3}" \
-            --argjson dry false \
-            '{
-              target_tag: $tag,
-              soak_seconds: $soak,
-              batch_size: $batch,
-              dry_run: $dry
-            }')
-
-          if [ -n "${CANARY_SLUG:-}" ]; then
-            BODY=$(jq '. * {canary_slug: $slug}' --arg slug "$CANARY_SLUG" <<<"$BODY")
-          fi
-
-          echo "Calling: POST $CP_URL/cp/admin/tenants/redeploy-fleet"
-          echo "  target_tag: $TARGET_TAG"
-          echo "  body: $BODY"
-
-          HTTP_RESPONSE=$(mktemp)
-          HTTP_CODE_FILE=$(mktemp)
-          set +e
-          curl -sS -o "$HTTP_RESPONSE" -w '%{http_code}' \
-            -m 1200 \
-            -H "Authorization: Bearer $CP_ADMIN_API_TOKEN" \
-            -H "Content-Type: application/json" \
-            -X POST "$CP_URL/cp/admin/tenants/redeploy-fleet" \
-            -d "$BODY" >"$HTTP_CODE_FILE"
-          CURL_EXIT=$?
-          set -e
-
-          HTTP_CODE=$(cat "$HTTP_CODE_FILE" 2>/dev/null || echo "000")
-          [ -z "$HTTP_CODE" ] && HTTP_CODE="000"
-
-          echo "HTTP $HTTP_CODE (curl exit $CURL_EXIT)"
-          cat "$HTTP_RESPONSE" | jq . || cat "$HTTP_RESPONSE"
-
-          if [ "$HTTP_CODE" -ge 400 ]; then
-            echo "::error::CP redeploy-fleet returned HTTP $HTTP_CODE — refusing to proceed."
-            exit 1
-          fi
-
-      - name: Summary
-        run: |
-          {
-            echo "## Canary verified — :latest promoted via CP redeploy-fleet"
-            echo ""
-            echo "- **Target tag:** \`staging-${{ needs.canary-smoke.outputs.sha }}\`"
-            echo "- **Registry:** ECR (\`${TENANT_IMAGE_NAME}\`)"
-            echo "- **Canary slug:** \`${CANARY_SLUG:-<none>}\` (soak ${SOAK_SECONDS}s)"
-            echo "- **Batch size:** ${BATCH_SIZE:-3}"
-            echo ""
-            echo "CP redeploy-fleet is rolling out the verified image across the prod fleet."
-            echo "The fleet's 5-minute health-check loop will pick up the update automatically."
-          } >> "$GITHUB_STEP_SUMMARY"
@@ -1,39 +0,0 @@
-name: cascade-list-drift-gate
-
-# Structural gate: TEMPLATES list in publish-runtime.yml must match
-# manifest.json's workspace_templates exactly. Closes the recurrence
-# path of PR #2556 (the data fix) and is the first concrete deliverable
-# of RFC #388 PR-3.
-#
-# Why a gate, not just discipline: PR #2536 pruned the manifest, but the
-# cascade list wasn't updated for ~weeks before someone (PR #2556)
-# noticed during an unrelated audit. During that window, codex never
-# rebuilt on a runtime publish. A structural gate catches the drift
-# the same day either file changes.
-#
-# Triggers narrowly to keep CI quiet: only on PRs that actually change
-# one of the two files. The path-filtered split + always-emit-result
-# pattern (memory: "Required check names need a job that always runs")
-# is unnecessary here because the workflow IS the check name and PR
-# branch protection should require it directly. Future-proof: if this
-# becomes a required check, add a no-op aggregator with always() so the
-# name still emits when paths don't match.
-
-on:
-  pull_request:
-    branches: [staging, main]
-    paths:
-      - manifest.json
-      - .github/workflows/publish-runtime.yml
-      - scripts/check-cascade-list-vs-manifest.sh
-
-permissions:
-  contents: read
-
-jobs:
-  check:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
-      - name: Check cascade list matches manifest
-        run: bash scripts/check-cascade-list-vs-manifest.sh
@@ -1,58 +0,0 @@
-name: Check migration collisions
-
-# Hard gate (#2341): fails a PR that adds a migration prefix already
-# claimed by the base branch or another open PR. Caught manually 2026-04-30
-# during PR #2276 rebase: 044_runtime_image_pins collided with
-# 044_platform_inbound_secret from RFC #2312. This workflow makes that
-# check automatic.
-#
-# Trigger model: pull_request only — there's no value running this on
-# pushes to staging or main (those are post-merge; the gate must fire
-# pre-merge to be useful). Path filter scopes to PRs that actually touch
-# migrations.
-
-on:
-  pull_request:
-    types: [opened, synchronize, reopened]
-    paths:
-      - 'workspace-server/migrations/**'
-      - 'scripts/ops/check_migration_collisions.py'
-      - '.github/workflows/check-migration-collisions.yml'
-
-permissions:
-  contents: read
-  # gh pr list/diff need read access to other PRs
-  pull-requests: read
-
-jobs:
-  check:
-    name: Migration version collision check
-    runs-on: ubuntu-latest
-    timeout-minutes: 5
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-        with:
-          # Need history to diff against base ref
-          fetch-depth: 0
-
-      - name: Detect collisions
-        env:
-          PR_NUMBER: ${{ github.event.pull_request.number }}
-          BASE_REF: origin/${{ github.event.pull_request.base.ref }}
-          HEAD_REF: ${{ github.event.pull_request.head.sha }}
-          GITHUB_REPOSITORY: ${{ github.repository }}
-          # gh CLI uses GH_TOKEN from env
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          # Ensure the named base ref exists locally. checkout@v4 with
-          # fetch-depth=0 pulls full history, but the explicit fetch is
-          # cheap insurance against form-of-ref differences across runs.
-          #
-          # IMPORTANT: do NOT pass --depth=1 here. The script below uses
-          # `git diff origin/<base>...<head>` (three-dot, merge-base form),
-          # which fails with "fatal: no merge base" if the base ref is
-          # shallow. The auto-promote staging→main PR (#2361) was blocked
-          # by exactly this for ~5h on 2026-04-30 — the depth=1 fetch
-          # overwrote checkout@v4's full-history clone with a shallow tip.
-          git fetch origin "${{ github.event.pull_request.base.ref }}" || true
-          python3 scripts/ops/check_migration_collisions.py
@@ -1,443 +0,0 @@
-name: CI
-
-on:
-  push:
-    branches: [main, staging]
-  pull_request:
-    branches: [main, staging]
-  # GitHub merge queue fires `merge_group` for the queue's pre-merge CI run.
-  # Required so the queue gets a real check result instead of a false-green
-  # from the absence of a triggered workflow. Safe to add unconditionally —
-  # the event simply doesn't fire until the queue is enabled on the branch.
-  merge_group:
-    types: [checks_requested]
-
-# Cancel in-progress CI runs when a new commit arrives on the same ref.
-# This prevents stale runs from queuing behind each other. The merge_group
-# refs (refs/heads/gh-readonly-queue/...) get their own concurrency group
-# automatically because github.ref differs from the PR ref.
-concurrency:
-  group: ci-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  # Detect which paths changed so downstream jobs can skip when only
-  # docs/markdown files were modified.
-  changes:
-    name: Detect changes
-    runs-on: ubuntu-latest
-    outputs:
-      platform: ${{ steps.check.outputs.platform }}
-      canvas: ${{ steps.check.outputs.canvas }}
-      python: ${{ steps.check.outputs.python }}
-      scripts: ${{ steps.check.outputs.scripts }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-        with:
-          fetch-depth: 0
-      - id: check
-        run: |
-          # For PR events: diff against the base branch (not HEAD~1 of the branch,
-          # which may be unrelated after force-pushes). When a push updates a PR,
-          # both pull_request and push events fire — prefer the PR base so that
-          # the diff is always computed against the actual merge base, not the
-          # previous SHA on the branch which may be on a different history line.
-          BASE="${GITHUB_BASE_REF:-${{ github.event.before }}}"
-          # GITHUB_BASE_REF is set by GitHub for PR events (the base branch name).
-          # For pull_request events we use the stored base.sha; for push events
-          # (or when base.sha is unavailable) fall back to github.event.before.
-          if [ "${{ github.event_name }}" = "pull_request" ] && [ -n "${{ github.event.pull_request.base.sha }}" ]; then
-            BASE="${{ github.event.pull_request.base.sha }}"
-          fi
-          # Fallback: if BASE is empty or all zeros (new branch), run everything
-          if [ -z "$BASE" ] || echo "$BASE" | grep -qE '^0+$'; then
-            echo "platform=true" >> "$GITHUB_OUTPUT"
-            echo "canvas=true" >> "$GITHUB_OUTPUT"
-            echo "python=true" >> "$GITHUB_OUTPUT"
-            echo "scripts=true" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-          DIFF=$(git diff --name-only "$BASE" HEAD 2>/dev/null || echo ".github/workflows/ci.yml")
-          echo "platform=$(echo "$DIFF" | grep -qE '^workspace-server/|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "canvas=$(echo "$DIFF" | grep -qE '^canvas/|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "python=$(echo "$DIFF" | grep -qE '^workspace/|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "scripts=$(echo "$DIFF" | grep -qE '^tests/e2e/|^scripts/|^infra/scripts/|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-
-  # Platform (Go) is a required check on staging. Always-run + per-step
-  # gating (see Canvas (Next.js) for the rationale and the failure mode
-  # this avoids).
-  platform-build:
-    name: Platform (Go)
-    needs: changes
-    runs-on: ubuntu-latest
-    defaults:
-      run:
-        working-directory: workspace-server
-    steps:
-      - if: needs.changes.outputs.platform != 'true'
-        working-directory: .
-        run: echo "No platform/** changes — skipping real build steps; this job always runs to satisfy the required-check name on branch protection."
-      - if: needs.changes.outputs.platform == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.platform == 'true'
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
-        with:
-          go-version: 'stable'
-      - if: needs.changes.outputs.platform == 'true'
-        run: go mod download
-      - if: needs.changes.outputs.platform == 'true'
-        run: go build ./cmd/server
-      # CLI (molecli) moved to standalone repo: github.com/molecule-ai/molecule-cli
-      - if: needs.changes.outputs.platform == 'true'
-        run: go vet ./... || true
-      - if: needs.changes.outputs.platform == 'true'
-        name: Run golangci-lint
-        run: golangci-lint run --timeout 3m ./... || true
-      - if: needs.changes.outputs.platform == 'true'
-        name: Run tests with race detection and coverage
-        run: go test -race -coverprofile=coverage.out ./...
-
-      - if: needs.changes.outputs.platform == 'true'
-        name: Per-file coverage report
-        # Advisory — lists every source file with its coverage so reviewers
-        # can see at-a-glance where gaps are. Sorted ascending so the worst
-        # offenders float to the top. Does NOT fail the build; the hard
-        # gate is the threshold check below. (#1823)
-        run: |
-          echo "=== Per-file coverage (worst first) ==="
-          go tool cover -func=coverage.out \
-            | grep -v '^total:' \
-            | awk '{file=$1; sub(/:[0-9][0-9.]*:.*/, "", file); pct=$NF; gsub(/%/,"",pct); s[file]+=pct; c[file]++}
-                   END {for (f in s) printf "%6.1f%%  %s\n", s[f]/c[f], f}' \
-            | sort -n
-
-      - if: needs.changes.outputs.platform == 'true'
-        name: Check coverage thresholds
-        # Enforces two gates from #1823 Layer 1:
-        #   1. Total floor (25% — ratchet plan in COVERAGE_FLOOR.md).
-        #   2. Per-file floor — non-test .go files in security-critical
-        #      paths with coverage <10% fail the build, UNLESS the file
-        #      path is listed in .coverage-allowlist.txt (acknowledged
-        #      historical debt with a tracking issue + expiry).
-        run: |
-          set -e
-          TOTAL_FLOOR=25
-          # Security-critical paths where a 0%-coverage file is a real risk.
-          CRITICAL_PATHS=(
-            "internal/handlers/tokens"
-            "internal/handlers/workspace_provision"
-            "internal/handlers/a2a_proxy"
-            "internal/handlers/registry"
-            "internal/handlers/secrets"
-            "internal/middleware/wsauth"
-            "internal/crypto"
-          )
-
-          TOTAL=$(go tool cover -func=coverage.out | grep '^total:' | awk '{print $3}' | sed 's/%//')
-          echo "Total coverage: ${TOTAL}%"
-          if awk "BEGIN{exit !($TOTAL < $TOTAL_FLOOR)}"; then
-            echo "::error::Total coverage ${TOTAL}% is below the ${TOTAL_FLOOR}% floor. See COVERAGE_FLOOR.md for ratchet plan."
-            exit 1
-          fi
-
-          # Aggregate per-file coverage → /tmp/perfile.txt: "<fullpath> <pct>"
-          go tool cover -func=coverage.out \
-            | grep -v '^total:' \
-            | awk '{file=$1; sub(/:[0-9][0-9.]*:.*/, "", file); pct=$NF; gsub(/%/,"",pct); s[file]+=pct; c[file]++}
-                   END {for (f in s) printf "%s %.1f\n", f, s[f]/c[f]}' \
-            > /tmp/perfile.txt
-
-          # Build allowlist — paths relative to workspace-server, one per line.
-          # Lines starting with # are comments.
-          ALLOWLIST=""
-          if [ -f ../.coverage-allowlist.txt ]; then
-            ALLOWLIST=$(grep -vE '^(#|[[:space:]]*$)' ../.coverage-allowlist.txt || true)
-          fi
-
-          FAILED=0
-          WARNED=0
-          for path in "${CRITICAL_PATHS[@]}"; do
-            while read -r file pct; do
-              [[ "$file" == *_test.go ]] && continue
-              [[ "$file" == *"$path"* ]] || continue
-              awk "BEGIN{exit !($pct < 10)}" || continue
-
-              # Strip the package-import prefix so we can match .coverage-allowlist.txt
-              # entries written as paths relative to workspace-server/.
-              # Handle both module paths: platform/workspace-server/... and platform/...
-              rel=$(echo "$file" | sed 's|^github.com/molecule-ai/molecule-monorepo/platform/workspace-server/||; s|^github.com/molecule-ai/molecule-monorepo/platform/||')
-
-              if echo "$ALLOWLIST" | grep -qxF "$rel"; then
-                echo "::warning file=workspace-server/$rel::Critical file at ${pct}% coverage (allowlisted, #1823) — fix before expiry."
-                WARNED=$((WARNED+1))
-              else
-                echo "::error file=workspace-server/$rel::Critical file at ${pct}% coverage — must be >=10% (target 80%). See #1823. To acknowledge as known debt, add this path to .coverage-allowlist.txt."
-                FAILED=$((FAILED+1))
-              fi
-            done < /tmp/perfile.txt
-          done
-
-          echo ""
-          echo "Critical-path check: $FAILED new failures, $WARNED allowlisted warnings."
-
-          if [ "$FAILED" -gt 0 ]; then
-            echo ""
-            echo "$FAILED security-critical file(s) have <10% test coverage and are"
-            echo "NOT in the allowlist. These paths handle auth, tokens, secrets, or"
-            echo "workspace provisioning — a 0% file here is the exact gap that let"
-            echo "CWE-22, CWE-78, KI-005 slip through in past incidents. Either:"
-            echo "  (a) add tests to raise coverage above 10%, or"
-            echo "  (b) add the path to .coverage-allowlist.txt with an expiry date"
-            echo "      and a tracking issue reference."
-            exit 1
-          fi
-
-  # Canvas (Next.js) — required check, always runs. See platform-build
-  # comment above for the rationale.
-  #
-  # Supersedes the canvas-build-noop pattern attempted in PR #2321: two
-  # jobs sharing `name:` doesn't actually satisfy branch protection
-  # because the SKIPPED check run sibling is treated as not-passed
-  # regardless of how many SUCCESS siblings it has. Verified empirically
-  # on PR #2314 — mergeStateStatus stayed BLOCKED until I collapsed to
-  # a single-job-with-conditional-steps shape.
-  canvas-build:
-    name: Canvas (Next.js)
-    needs: changes
-    runs-on: ubuntu-latest
-    defaults:
-      run:
-        working-directory: canvas
-    steps:
-      - if: needs.changes.outputs.canvas != 'true'
-        working-directory: .
-        run: echo "No canvas/** changes — skipping real build steps; this job always runs to satisfy the required-check name on branch protection."
-      - if: needs.changes.outputs.canvas == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.canvas == 'true'
-        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
-        with:
-          node-version: '22'
-      - if: needs.changes.outputs.canvas == 'true'
-        run: rm -f package-lock.json && npm install
-      - if: needs.changes.outputs.canvas == 'true'
-        run: npm run build
-      - if: needs.changes.outputs.canvas == 'true'
-        name: Run tests with coverage
-        # Coverage instrumentation is configured in canvas/vitest.config.ts
-        # (provider: v8, reporters: text + html + json-summary). Step 2 of
-        # #1815 — wires coverage into CI so we get a baseline visible on
-        # every PR. No threshold gate yet; thresholds dial in (Step 3, also
-        # tracked in #1815) after the team sees what current coverage is.
-        # Per the inline comment in vitest.config.ts: "first land
-        # observability so we can see the baseline, then dial in
-        # thresholds + a hard gate" — this PR ships the observability half.
-        run: npx vitest run --coverage
-      - name: Upload coverage summary as artifact
-        if: needs.changes.outputs.canvas == 'true' && always()
-        # Pinned to v3 for Gitea act_runner v0.6 compatibility — v4+ uses
-        # the GHES 3.10+ artifact protocol that Gitea 1.22.x does NOT
-        # implement, surfacing as `GHESNotSupportedError: @actions/artifact
-        # v2.0.0+, upload-artifact@v4+ and download-artifact@v4+ are not
-        # currently supported on GHES`. Drop this pin when Gitea ships
-        # the v4 protocol (tracked: post-Gitea-1.23 followup).
-        uses: actions/upload-artifact@c6a366c94c3e0affe28c06c8df20a878f24da3cf # v3.2.2
-        with:
-          name: canvas-coverage-${{ github.run_id }}
-          path: canvas/coverage/
-          retention-days: 7
-          if-no-files-found: warn
-
-  # MCP Server + SDK removed from CI — now in standalone repos:
-  # - github.com/molecule-ai/molecule-mcp-server (npm CI)
-  # - github.com/molecule-ai/molecule-sdk-python (PyPI CI)
-
-  # e2e-api job moved to .github/workflows/e2e-api.yml (issue #458).
-  # It now has workflow-level concurrency (cancel-in-progress: false) so
-  # new pushes queue the E2E run rather than cancelling it at the run level.
-
-  # Shellcheck (E2E scripts) — required check, always runs. See
-  # platform-build for the rationale.
-  shellcheck:
-    name: Shellcheck (E2E scripts)
-    needs: changes
-    runs-on: ubuntu-latest
-    steps:
-      - if: needs.changes.outputs.scripts != 'true'
-        run: echo "No tests/e2e/ or infra/scripts/ changes — skipping real shellcheck; this job always runs to satisfy the required-check name on branch protection."
-      - if: needs.changes.outputs.scripts == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.scripts == 'true'
-        name: Run shellcheck on tests/e2e/*.sh and infra/scripts/*.sh
-        # shellcheck is pre-installed on ubuntu-latest runners (via apt).
-        # infra/scripts/ is included because setup.sh + nuke.sh gate the
-        # README quickstart — a shellcheck regression there silently breaks
-        # new-user onboarding. scripts/ is intentionally excluded until its
-        # pre-existing SC3040/SC3043 warnings are cleaned up.
-        run: |
-          find tests/e2e infra/scripts -type f -name '*.sh' -print0 \
-            | xargs -0 shellcheck --severity=warning
-
-      - if: needs.changes.outputs.scripts == 'true'
-        name: Lint cleanup-trap hygiene (RFC #2873)
-        # Asserts every shell E2E test that calls `mktemp` also installs
-        # an EXIT trap. Catches the /tmp-leak class — a missing trap
-        # silently leaks scratch into CI runners (~10-100KB per run).
-        # See tests/e2e/lint_cleanup_traps.sh for the rule + fix pattern.
-        run: bash tests/e2e/lint_cleanup_traps.sh
-
-      - if: needs.changes.outputs.scripts == 'true'
-        name: Run E2E bash unit tests (no live infra)
-        # Pure-bash unit tests for E2E helper libs (lib/*.sh). These pin
-        # behavior of dispatch logic that — when broken — silently masks as
-        # "Could not resolve authentication method" only after a successful
-        # tenant + workspace provision (PR #2571 incident, 2026-05-03). Add
-        # new self-contained unit tests here as the lib/ directory grows;
-        # tests requiring live CP/tenant credentials belong in the dedicated
-        # e2e-staging-* workflows, not this job.
-        run: |
-          bash tests/e2e/test_model_slug.sh
-
-  canvas-deploy-reminder:
-    name: Canvas Deploy Reminder
-    runs-on: ubuntu-latest
-    needs: [changes, canvas-build]
-    # Only fires on direct pushes to main (i.e. after staging→main promotion).
-    if: needs.changes.outputs.canvas == 'true' && github.event_name == 'push' && github.ref == 'refs/heads/main'
-    steps:
-      - name: Write deploy reminder to step summary
-        env:
-          COMMIT_SHA: ${{ github.sha }}
-          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
-        run: |
-          # Write body to a temp file — avoids backtick escaping in shell.
-          cat > /tmp/deploy-reminder.md << 'BODY'
-          ## Canvas build passed ✅ — deploy required
-
-          The `publish-canvas-image` workflow is now building a fresh Docker image
-          (`ghcr.io/molecule-ai/canvas:latest`) in the background.
-
-          Once it completes (~3–5 min), apply on the host machine with:
-          ```bash
-          cd <runner-workspace>
-          git pull origin main
-          docker compose pull canvas && docker compose up -d canvas
-          ```
-
-          If you need to rebuild from local source instead (e.g. testing unreleased
-          changes or a new `NEXT_PUBLIC_*` URL), use:
-          ```bash
-          docker compose build canvas && docker compose up -d canvas
-          ```
-          BODY
-          printf '\n> Posted automatically by CI · commit `%s` · [build log](%s)\n' \
-            "$COMMIT_SHA" "$RUN_URL" >> /tmp/deploy-reminder.md
-
-          # Gitea has no commit-comments API (no equivalent of
-          # POST /repos/{owner}/{repo}/commits/{commit_sha}/comments).
-          # Write to GITHUB_STEP_SUMMARY instead — both GitHub Actions and
-          # Gitea Actions render this as the workflow run's summary page,
-          # which is where operators look for post-deploy action items.
-          # (#75 / PR-D)
-          cat /tmp/deploy-reminder.md >> "$GITHUB_STEP_SUMMARY"
-
-  # Python Lint & Test — required check, always runs. See platform-build
-  # for the rationale.
-  python-lint:
-    name: Python Lint & Test
-    needs: changes
-    runs-on: ubuntu-latest
-    env:
-      WORKSPACE_ID: test
-    defaults:
-      run:
-        working-directory: workspace
-    steps:
-      - if: needs.changes.outputs.python != 'true'
-        working-directory: .
-        run: echo "No workspace/** changes — skipping real lint+test; this job always runs to satisfy the required-check name on branch protection."
-      - if: needs.changes.outputs.python == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.python == 'true'
-        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
-        with:
-          python-version: '3.11'
-          cache: pip
-          cache-dependency-path: workspace/requirements.txt
-      - if: needs.changes.outputs.python == 'true'
-        run: pip install -r requirements.txt pytest pytest-asyncio pytest-cov sqlalchemy>=2.0.0
-      # Coverage flags + fail-under floor moved into workspace/pytest.ini
-      # (issue #1817) so local `pytest` and CI use identical config.
-      - if: needs.changes.outputs.python == 'true'
-        run: python -m pytest --tb=short
-
-      - if: needs.changes.outputs.python == 'true'
-        name: Per-file critical-path coverage (MCP / inbox / auth)
-        # MCP-critical Python files have a per-file floor on top of the
-        # 86% total floor in pytest.ini. Rationale (issue #2790, after
-        # the PR #2766 → PR #2771 cycle): the total floor averages ~6000
-        # lines, so a single MCP file could regress to ~50% with no
-        # complaint as long as other modules compensate. These five
-        # files handle multi-tenant routing + auth + inbox dispatch —
-        # a coverage drop here is the same risk shape as a Go-side
-        # workspace-server token/secrets file dropping below 10%.
-        #
-        # Floor 75% sits below current actuals (80-96%) so this gate is
-        # strictly additive — no existing PR fails. Ratchet plan in
-        # COVERAGE_FLOOR.md.
-        run: |
-          set -e
-          PER_FILE_FLOOR=75
-          CRITICAL_FILES=(
-            "a2a_mcp_server.py"
-            "mcp_cli.py"
-            "a2a_tools.py"
-            "a2a_tools_inbox.py"
-            "inbox.py"
-            "platform_auth.py"
-          )
-
-          # pytest already wrote .coverage; emit a JSON view scoped to
-          # the critical files so jq/python can read the per-file pct
-          # without parsing tabular text. --include uses fnmatch, and
-          # the leading "*" allows the file to live anywhere under the
-          # workspace root (today they sit at workspace/<name>.py).
-          INCLUDES=$(printf '*%s,' "${CRITICAL_FILES[@]}")
-          INCLUDES="${INCLUDES%,}"
-          python -m coverage json -o /tmp/critical-cov.json --include="$INCLUDES"
-
-          FAILED=0
-          for f in "${CRITICAL_FILES[@]}"; do
-            # Match by top-level path key (e.g. "a2a_tools.py", not
-            # "builtin_tools/a2a_tools.py" — different file at 100%).
-            # The keys in coverage.json are paths relative to the run
-            # cwd (workspace/), so the critical-path entry sits at the
-            # bare basename.
-            pct=$(jq -r --arg f "$f" '.files | to_entries | map(select(.key == $f)) | .[0].value.summary.percent_covered // "MISSING"' /tmp/critical-cov.json)
-            if [ "$pct" = "MISSING" ]; then
-              echo "::error file=workspace/$f::No coverage data — file may have moved or test exclusion mis-set."
-              FAILED=$((FAILED+1))
-              continue
-            fi
-            echo "$f: ${pct}%"
-            if awk "BEGIN{exit !($pct < $PER_FILE_FLOOR)}"; then
-              echo "::error file=workspace/$f::${pct}% < ${PER_FILE_FLOOR}% per-file floor (MCP critical path). See COVERAGE_FLOOR.md."
-              FAILED=$((FAILED+1))
-            fi
-          done
-
-          if [ "$FAILED" -gt 0 ]; then
-            echo ""
-            echo "$FAILED MCP critical-path file(s) below the ${PER_FILE_FLOOR}% per-file floor."
-            echo "These paths handle multi-tenant routing, auth tokens, and inbox dispatch."
-            echo "A coverage drop here is the same risk shape as Go-side tokens/secrets files"
-            echo "dropping below 10% (see COVERAGE_FLOOR.md). Either:"
-            echo "  (a) add tests to raise coverage back above ${PER_FILE_FLOOR}%, or"
-            echo "  (b) if this is unavoidable historical debt, file an issue and propose"
-            echo "      adjusting the floor with rationale in COVERAGE_FLOOR.md."
-            exit 1
-          fi
-
-      # SDK + plugin validation moved to standalone repo:
-      # github.com/molecule-ai/molecule-sdk-python
-
@@ -1,257 +0,0 @@
-name: Continuous synthetic E2E (staging)
-
-# Hard gate (#2342): cron-driven full-lifecycle E2E that catches
-# regressions visible only at runtime — schema drift, deployment-pipeline
-# gaps, vendor outages, env-var rotations, DNS / CF / Railway side-effects.
-#
-# Why this gate exists:
-#   PR-time CI catches code-level regressions but not deployment-time or
-#   integration-time ones. Today's empirical data:
-#     • #2345 (A2A v0.2 silent drop) — passed all unit tests, broke at
-#       JSON-RPC parse layer between sender and receiver. Visible only
-#       to a sender exercising the full path.
-#     • RFC #2312 chat upload — landed on staging-branch but never
-#       reached staging tenants because publish-workspace-server-image
-#       was main-only. Caught by manual dogfooding hours after deploy.
-#   Both would have surfaced within 15-20 min of regression if a
-#   continuous synth-E2E was running.
-#
-# Cadence: every 20 min (3x/hour). The script is conservatively
-# bounded at 10 min wall-clock; even on degraded staging it should
-# finish before the next firing. cron-overlap is guarded by the
-# concurrency group below.
-#
-# Cost: ~3 runs/hour × 5-10 min × $0.008/min GHA = ~$0.50-$1/day.
-# Plus a fresh tenant provisioned + torn down each run (Railway +
-# AWS pennies). Negligible.
-#
-# Failure handling: when the run fails, the workflow exits non-zero
-# and GitHub's standard email/notification path fires. Operators
-# can subscribe to this workflow's failure channel for paging-grade
-# alerting.
-
-on:
-  schedule:
-    # Every 10 minutes, on :02 :12 :22 :32 :42 :52. Three constraints:
-    #   1. Stay off the top-of-hour. GitHub Actions scheduler drops
-    #      :00 firings under high load (own docs:
-    #      https://docs.github.com/en/actions/using-workflows/events-that-trigger-workflows#schedule).
-    #      Prior history: cron was '0,20,40' (2026-05-02) — only :00
-    #      ever survived. Bumped to '10,30,50' (2026-05-03) on the
-    #      theory that further-from-:00 wins. Empirically 2026-05-04
-    #      that ALSO dropped to ~60 min effective cadence (only ~1
-    #      schedule fire per hour — see molecule-core#2726). Detection
-    #      latency was claimed 20 min, actual 60 min.
-    #   2. Avoid colliding with the existing :15 sweep-cf-orphans
-    #      and :45 sweep-cf-tunnels — both hit the CF API and we
-    #      don't want to fight for rate-limit tokens.
-    #   3. Avoid the :30 heavy slot (canary-staging /30, sweep-aws-
-    #      secrets, sweep-stale-e2e-orgs every :15) — multiple
-    #      overlapping cron registrations on the same minute is part
-    #      of what GH drops under load.
-    # Solution: bump fires-per-hour 3 → 6 AND keep all slots in clean
-    # lanes (1-3 min away from any other cron). Even with empirically-
-    # observed ~67% GH drop ratio, 6 attempts/hour yields ~2 effective
-    # fires = ~30 min cadence; closer to the 20-min target than the
-    # current shape and provides a real degradation alarm if drops
-    # get worse.
-    - cron: '2,12,22,32,42,52 * * * *'
-  workflow_dispatch:
-    inputs:
-      runtime:
-        description: "Runtime to provision (claude-code = default + cheapest via MiniMax; langgraph = OpenAI-only; hermes = SDK-native path, slower)"
-        required: false
-        default: "claude-code"
-        type: string
-      model_slug:
-        description: "Model id to provision the workspace with (default MiniMax-M2.7-highspeed; e.g. 'sonnet' to test direct Anthropic, 'openai/gpt-4o' for hermes)"
-        required: false
-        default: "MiniMax-M2.7-highspeed"
-        type: string
-      keep_org:
-        description: "Skip teardown for post-mortem debugging (only manual dispatch — never set this for cron runs)"
-        required: false
-        default: false
-        type: boolean
-
-permissions:
-  contents: read
-  # No issue-write here — failures surface as red runs in the workflow
-  # history. If you want auto-issue-on-fail, add a follow-up step that
-  # uses gh issue create gated on `if: failure()`. Keeping the surface
-  # minimal until that's actually wanted.
-
-# Serialize so two firings can never overlap. Cron firing every 20 min
-# but scripts conservatively bounded at 10 min — overlap shouldn't
-# happen in steady state, but if a run hangs we don't want N more
-# stacking up.
-concurrency:
-  group: continuous-synth-e2e
-  cancel-in-progress: false
-
-jobs:
-  synth:
-    name: Synthetic E2E against staging
-    runs-on: ubuntu-latest
-    # Bumped from 12 → 20 (2026-05-04). Tenant user-data install phase
-    # (apt-get update + install docker.io/jq/awscli/caddy + snap install
-    # ssm-agent) runs from raw Ubuntu on every boot — none of it is
-    # pre-baked into the tenant AMI. Empirical fetch_secrets/ok timing
-    # across today's canaries: 51s → 82s → 143s → 625s. apt-mirror tail
-    # latency drives the boot-to-fetch_secrets phase from ~1min to >10min.
-    # A 12min budget leaves only ~2min for the workspace (which needs
-    # ~3.5min for claude-code cold boot) on slow-apt days, blowing the
-    # budget. 20min absorbs the worst tenant tail so the workspace probe
-    # gets the full ~7min it needs even on a slow apt day. Real fix:
-    # pre-bake caddy + ssm-agent into the tenant AMI (controlplane#TBD).
-    timeout-minutes: 20
-    env:
-      # claude-code default: cold-start ~5 min (comparable to langgraph),
-      # but uses MiniMax-M2.7-highspeed via the template's third-party-
-      # Anthropic-compat path (workspace-configs-templates/claude-code-
-      # default/config.yaml:64-69). MiniMax is ~5-10x cheaper than
-      # gpt-4.1-mini per token AND avoids the recurring OpenAI quota-
-      # exhaustion class that took the canary down 2026-05-03 (#265).
-      # Operators can pick langgraph / hermes via workflow_dispatch
-      # when they specifically need to exercise the OpenAI or SDK-
-      # native paths.
-      E2E_RUNTIME: ${{ github.event.inputs.runtime || 'claude-code' }}
-      # Pin the canary to a specific MiniMax model rather than relying
-      # on the per-runtime default ("sonnet" → routes to direct
-      # Anthropic, defeats the cost saving). Operators can override
-      # via workflow_dispatch by setting a different E2E_MODEL_SLUG
-      # input if they need to exercise a specific model. M2.7-highspeed
-      # is "Token Plan only" but cheap-per-token and fast.
-      E2E_MODEL_SLUG: ${{ github.event.inputs.model_slug || 'MiniMax-M2.7-highspeed' }}
-      # Bound to 10 min so a stuck provision fails the run instead of
-      # holding up the next cron firing. 15-min default in the script
-      # is for the on-PR full lifecycle where we have more headroom.
-      E2E_PROVISION_TIMEOUT_SECS: '600'
-      # Slug suffix — namespaced "synth-" so these runs are
-      # distinguishable from PR-driven runs in CP admin.
-      E2E_RUN_ID: synth-${{ github.run_id }}
-      # Forced false for cron; respected for manual dispatch
-      E2E_KEEP_ORG: ${{ github.event.inputs.keep_org == 'true' && '1' || '' }}
-      MOLECULE_CP_URL: ${{ vars.STAGING_CP_URL || 'https://staging-api.moleculesai.app' }}
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_API_TOKEN }}
-      # MiniMax key is the canary's PRIMARY auth path. claude-code
-      # template's `minimax` provider routes ANTHROPIC_BASE_URL to
-      # api.minimax.io/anthropic and reads MINIMAX_API_KEY at boot.
-      # tests/e2e/test_staging_full_saas.sh branches SECRETS_JSON on
-      # which key is present — MiniMax wins when set.
-      E2E_MINIMAX_API_KEY: ${{ secrets.MOLECULE_STAGING_MINIMAX_API_KEY }}
-      # Direct-Anthropic alternative for operators who don't want to
-      # set up a MiniMax account (priority below MiniMax — first
-      # non-empty wins in test_staging_full_saas.sh's secrets-injection
-      # block). See #2578 PR comment for the rationale.
-      E2E_ANTHROPIC_API_KEY: ${{ secrets.MOLECULE_STAGING_ANTHROPIC_API_KEY }}
-      # OpenAI fallback — kept wired so operators can dispatch with
-      # E2E_RUNTIME=langgraph or =hermes and still have a working
-      # canary path. The script picks the right blob shape based on
-      # which key is non-empty.
-      E2E_OPENAI_API_KEY: ${{ secrets.MOLECULE_STAGING_OPENAI_KEY }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify required secrets present
-        run: |
-          # Hard-fail on missing secret REGARDLESS of trigger. Previously
-          # this step soft-skipped on workflow_dispatch via `exit 0`, but
-          # `exit 0` only ends the STEP — subsequent steps still ran with
-          # the empty secret, the synth script fell through to the wrong
-          # SECRETS_JSON branch, and the canary failed 5 min later with a
-          # confusing "Agent error (Exception)" instead of the clean
-          # "secret missing" message at the top. Caught 2026-05-04 by
-          # dispatched run 25296530706: claude-code + missing MINIMAX
-          # silently used OpenAI keys but kept model=MiniMax-M2.7, then
-          # the workspace 401'd against MiniMax once it tried to call.
-          # Fix: exit 1 in both cron and dispatch paths. Operators who
-          # want to verify a YAML change without setting up the secret
-          # can read the verify-secrets step's stderr — the failure is
-          # itself the verification signal.
-          if [ -z "${MOLECULE_ADMIN_TOKEN:-}" ]; then
-            echo "::error::CP_STAGING_ADMIN_API_TOKEN secret missing — synth E2E cannot run"
-            echo "::error::Set it at Settings → Secrets and Variables → Actions; pull from staging-CP's CP_ADMIN_API_TOKEN env in Railway."
-            exit 1
-          fi
-
-          # LLM-key requirement is per-runtime: claude-code accepts
-          # EITHER MiniMax OR direct-Anthropic (whichever is set first),
-          # langgraph + hermes use OpenAI (MOLECULE_STAGING_OPENAI_KEY).
-          case "${E2E_RUNTIME}" in
-            claude-code)
-              if [ -n "${E2E_MINIMAX_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY"
-                required_secret_value="${E2E_MINIMAX_API_KEY}"
-              elif [ -n "${E2E_ANTHROPIC_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value="${E2E_ANTHROPIC_API_KEY}"
-              else
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY or MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value=""
-              fi
-              ;;
-            langgraph|hermes)
-              required_secret_name="MOLECULE_STAGING_OPENAI_KEY"
-              required_secret_value="${E2E_OPENAI_API_KEY:-}"
-              ;;
-            *)
-              echo "::warning::Unknown E2E_RUNTIME='${E2E_RUNTIME}' — skipping LLM-key check"
-              required_secret_name=""
-              required_secret_value="present"
-              ;;
-          esac
-          if [ -n "$required_secret_name" ] && [ -z "$required_secret_value" ]; then
-            echo "::error::${required_secret_name} secret missing — runtime=${E2E_RUNTIME} cannot authenticate against its LLM provider"
-            echo "::error::Set it at Settings → Secrets and Variables → Actions, OR dispatch with a different runtime"
-            exit 1
-          fi
-
-      - name: Install required tools
-        run: |
-          # The script depends on jq + curl (already on ubuntu-latest)
-          # and python3 (likewise). Verify they're all present so we
-          # fail fast on a runner image regression rather than mid-script.
-          for cmd in jq curl python3; do
-            command -v "$cmd" >/dev/null 2>&1 || {
-              echo "::error::required tool '$cmd' not on PATH — runner image regression?"
-              exit 1
-            }
-          done
-
-      - name: Run synthetic E2E
-        # The script handles its own teardown via EXIT trap; even on
-        # failure (timeout, assertion), the org is deprovisioned and
-        # leaks are reported. Exit code propagates from the script.
-        run: |
-          bash tests/e2e/test_staging_full_saas.sh
-
-      - name: Failure summary
-        # Runs only on failure. Adds a job summary so the workflow run
-        # page shows a quick "what happened" instead of forcing readers
-        # to scroll through script output.
-        if: failure()
-        run: |
-          {
-            echo "## Continuous synth E2E failed"
-            echo ""
-            echo "**Run ID:** ${{ github.run_id }}"
-            echo "**Trigger:** ${{ github.event_name }}"
-            echo "**Runtime:** ${E2E_RUNTIME}"
-            echo "**Slug:** synth-${{ github.run_id }}"
-            echo ""
-            echo "### What this means"
-            echo ""
-            echo "Staging just regressed on a path that previously worked. Likely classes:"
-            echo "- Schema mismatch between sender and receiver (#2345 class)"
-            echo "- Deployment-pipeline gap (RFC #2312 / staging-tenant-image-stale class)"
-            echo "- Vendor outage (Cloudflare, Railway, AWS, GHCR)"
-            echo "- Staging-CP env var rotation"
-            echo ""
-            echo "### Next steps"
-            echo ""
-            echo "1. Check the script output above for the assertion that failed"
-            echo "2. If it's a vendor outage, no action needed — next firing in ~20 min"
-            echo "3. If it's a code regression, find the causing PR via \`git log\` against last green run and revert/fix"
-            echo "4. Keep an eye on the next 1-2 firings — flake vs persistent fail differs in priority"
-          } >> "$GITHUB_STEP_SUMMARY"
@@ -1,307 +0,0 @@
-name: E2E API Smoke Test
-# Extracted from ci.yml so workflow-level concurrency can protect this job
-# from run-level cancellation (issue #458).
-#
-# Trigger model (revised 2026-04-29):
-#
-# Always FIRES on push/pull_request to staging+main. Real work is gated
-# per-step on `needs.detect-changes.outputs.api` — when paths under
-# `workspace-server/`, `tests/e2e/`, or this workflow file haven't
-# changed, the no-op step alone runs and emits SUCCESS for the
-# `E2E API Smoke Test` check, satisfying branch protection without
-# spending CI cycles. See the in-job comment on the `e2e-api` job for
-# why this is one job (not two-jobs-sharing-name) and the 2026-04-29
-# PR #2264 incident that drove the consolidation.
-#
-# Parallel-safety (Class B Hongming-owned CICD red sweep, 2026-05-08)
-# -------------------------------------------------------------------
-# Same substrate hazard as PR #98 (handlers-postgres-integration). Our
-# Gitea act_runner runs with `container.network: host` (operator host
-# `/opt/molecule/runners/config.yaml`), which means:
-#
-#   * Two concurrent runs both try to bind their `-p 15432:5432` /
-#     `-p 16379:6379` host ports — the second postgres/redis FATALs
-#     with `Address in use` and `docker run` returns exit 125 with
-#     `Conflict. The container name "/molecule-ci-postgres" is already
-#     in use by container ...`. Verified in run a7/2727 on 2026-05-07.
-#   * The fixed container names `molecule-ci-postgres` / `-redis` (the
-#     pre-fix shape) collide on name AS WELL AS port. The cleanup-with-
-#     `docker rm -f` at the start of the second job KILLS the first
-#     job's still-running postgres/redis.
-#
-# Fix shape (mirrors PR #98's bridge-net pattern, adapted because
-# platform-server is a Go binary on the host, not a containerised
-# step):
-#
-#   1. Unique container names per run:
-#         pg-e2e-api-${RUN_ID}-${RUN_ATTEMPT}
-#         redis-e2e-api-${RUN_ID}-${RUN_ATTEMPT}
-#      `${RUN_ID}-${RUN_ATTEMPT}` is unique even across reruns of the
-#      same run_id.
-#   2. Ephemeral host port per run (`-p 0:5432`), then read the actual
-#      bound port via `docker port` and export DATABASE_URL/REDIS_URL
-#      pointing at it. No fixed host-port → no port collision.
-#   3. `127.0.0.1` (NOT `localhost`) in URLs — IPv6 first-resolve was
-#      the original flake fixed in #92 and the script's still IPv6-
-#      enabled.
-#   4. `if: always()` cleanup so containers don't leak when test steps
-#      fail.
-#
-# Issue #94 items #2 + #3 (also fixed here):
-#   * Pre-pull `alpine:latest` so the platform-server's provisioner
-#     (`internal/handlers/container_files.go`) can stand up its
-#     ephemeral token-write helper without a daemon.io round-trip.
-#   * Create `molecule-core-net` bridge network if missing so the
-#     provisioner's container.HostConfig {NetworkMode: ...} attach
-#     succeeds.
-# Item #1 (timeouts) — evidence on recent runs (77/3191, ae/4270, 0e/
-# 2318) shows Postgres ready in 3s, Redis in 1s, Platform in 1s when
-# they DO come up. Timeouts are not the bottleneck; not bumped.
-#
-# Item explicitly NOT fixed here: failing test `Status back online`
-# fails because the platform's langgraph workspace template image
-# (ghcr.io/molecule-ai/workspace-template-langgraph:latest) returns
-# 403 Forbidden post-2026-05-06 GitHub org suspension. That is a
-# template-registry resolution issue (ADR-002 / local-build mode) and
-# belongs in a separate change that touches workspace-server, not
-# this workflow file.
-
-on:
-  push:
-    branches: [main, staging]
-  pull_request:
-    branches: [main, staging]
-  workflow_dispatch:
-
-concurrency:
-  # Per-SHA grouping (changed 2026-04-28 from per-ref). Per-ref had the
-  # same auto-promote-staging brittleness as e2e-staging-canvas — back-
-  # to-back staging pushes share refs/heads/staging, so the older push's
-  # queued run gets cancelled when a newer push lands. Auto-promote-
-  # staging then sees `completed/cancelled` for the older SHA and stays
-  # put; the newer SHA's gates may eventually save the day, but if the
-  # newer push gets cancelled too, we deadlock.
-  #
-  # See e2e-staging-canvas.yml's identical concurrency block for the full
-  # rationale and the 2026-04-28 incident reference.
-  group: e2e-api-${{ github.event.pull_request.head.sha || github.sha }}
-  cancel-in-progress: false
-
-jobs:
-  detect-changes:
-    runs-on: ubuntu-latest
-    outputs:
-      api: ${{ steps.decide.outputs.api }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
-        id: filter
-        with:
-          filters: |
-            api:
-              - 'workspace-server/**'
-              - 'tests/e2e/**'
-              - '.github/workflows/e2e-api.yml'
-      - id: decide
-        # Always run real work for manual dispatch — no diff context to
-        # filter against and ops dispatching this expects the suite to
-        # actually exercise the platform.
-        run: |
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-            echo "api=true" >> "$GITHUB_OUTPUT"
-          else
-            echo "api=${{ steps.filter.outputs.api }}" >> "$GITHUB_OUTPUT"
-          fi
-
-  # ONE job (no job-level `if:`) that always runs and reports under the
-  # required-check name `E2E API Smoke Test`. Real work is gated per-step
-  # on `needs.detect-changes.outputs.api`. Reason: GitHub registers a
-  # check run for every job that matches `name:`, and a job-level
-  # `if: false` produces a SKIPPED check run. Branch protection treats
-  # all check runs with a matching context name on the latest commit as a
-  # SET — any SKIPPED in the set fails the required-check eval, even with
-  # SUCCESS siblings. Verified 2026-04-29 on PR #2264 (staging→main):
-  # 4 check runs (2 SKIPPED + 2 SUCCESS) at the head SHA blocked
-  # promotion despite all real work succeeding. Collapsing to a single
-  # always-running job with conditional steps emits exactly one SUCCESS
-  # check run regardless of paths filter — branch-protection-clean.
-  e2e-api:
-    needs: detect-changes
-    name: E2E API Smoke Test
-    runs-on: ubuntu-latest
-    timeout-minutes: 15
-    env:
-      # Unique per-run container names so concurrent runs on the host-
-      # network act_runner don't collide on name OR port.
-      # `${RUN_ID}-${RUN_ATTEMPT}` stays unique across reruns of the
-      # same run_id. PORT is set later (after docker port lookup) since
-      # we let Docker assign an ephemeral host port.
-      PG_CONTAINER: pg-e2e-api-${{ github.run_id }}-${{ github.run_attempt }}
-      REDIS_CONTAINER: redis-e2e-api-${{ github.run_id }}-${{ github.run_attempt }}
-      PORT: "8080"
-    steps:
-      - name: No-op pass (paths filter excluded this commit)
-        if: needs.detect-changes.outputs.api != 'true'
-        run: |
-          echo "No workspace-server / tests/e2e / workflow changes — E2E API gate satisfied without running tests."
-          echo "::notice::E2E API Smoke Test no-op pass (paths filter excluded this commit)."
-      - if: needs.detect-changes.outputs.api == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.detect-changes.outputs.api == 'true'
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
-        with:
-          go-version: 'stable'
-          cache: true
-          cache-dependency-path: workspace-server/go.sum
-      - name: Pre-pull alpine + ensure provisioner network (Issue #94 items #2 + #3)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: |
-          # Provisioner uses alpine:latest for ephemeral token-write
-          # containers (workspace-server/internal/handlers/container_files.go).
-          # Pre-pull so the first provision in test_api.sh doesn't race
-          # the daemon's pull cache. Idempotent — `docker pull` is a no-op
-          # when the image is already present.
-          docker pull alpine:latest >/dev/null
-          # Provisioner attaches workspace containers to
-          # molecule-core-net (workspace-server/internal/provisioner/
-          # provisioner.go::DefaultNetwork). The bridge already exists on
-          # the operator host's docker daemon — `network create` is
-          # idempotent via `|| true`.
-          docker network create molecule-core-net >/dev/null 2>&1 || true
-          echo "alpine:latest pre-pulled; molecule-core-net ensured."
-      - name: Start Postgres (docker)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: |
-          # Defensive cleanup — only matches THIS run's container name,
-          # so it cannot kill a sibling run's postgres. (Pre-fix the
-          # name was static and this rm hit other runs' containers.)
-          docker rm -f "$PG_CONTAINER" 2>/dev/null || true
-          # `-p 0:5432` requests an ephemeral host port; we read it back
-          # below and export DATABASE_URL.
-          docker run -d --name "$PG_CONTAINER" \
-            -e POSTGRES_USER=dev -e POSTGRES_PASSWORD=dev -e POSTGRES_DB=molecule \
-            -p 0:5432 postgres:16 >/dev/null
-          # Resolve the host-side port assignment. `docker port` prints
-          # `0.0.0.0:NNNN` (and on host-net runners may also print an
-          # IPv6 line — take the first IPv4 line).
-          PG_PORT=$(docker port "$PG_CONTAINER" 5432/tcp | awk -F: '/^0\.0\.0\.0:/ {print $2; exit}')
-          if [ -z "$PG_PORT" ]; then
-            # Fallback: any first line. Some Docker versions print only
-            # one line.
-            PG_PORT=$(docker port "$PG_CONTAINER" 5432/tcp | head -1 | awk -F: '{print $NF}')
-          fi
-          if [ -z "$PG_PORT" ]; then
-            echo "::error::Could not resolve host port for $PG_CONTAINER"
-            docker port "$PG_CONTAINER" 5432/tcp || true
-            docker logs "$PG_CONTAINER" || true
-            exit 1
-          fi
-          # 127.0.0.1 (NOT localhost) — IPv6 first-resolve flake (#92).
-          echo "PG_PORT=${PG_PORT}" >> "$GITHUB_ENV"
-          echo "DATABASE_URL=postgres://dev:dev@127.0.0.1:${PG_PORT}/molecule?sslmode=disable" >> "$GITHUB_ENV"
-          echo "Postgres host port: ${PG_PORT}"
-          for i in $(seq 1 30); do
-            if docker exec "$PG_CONTAINER" pg_isready -U dev >/dev/null 2>&1; then
-              echo "Postgres ready after ${i}s"
-              exit 0
-            fi
-            sleep 1
-          done
-          echo "::error::Postgres did not become ready in 30s"
-          docker logs "$PG_CONTAINER" || true
-          exit 1
-      - name: Start Redis (docker)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: |
-          docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true
-          docker run -d --name "$REDIS_CONTAINER" -p 0:6379 redis:7 >/dev/null
-          REDIS_PORT=$(docker port "$REDIS_CONTAINER" 6379/tcp | awk -F: '/^0\.0\.0\.0:/ {print $2; exit}')
-          if [ -z "$REDIS_PORT" ]; then
-            REDIS_PORT=$(docker port "$REDIS_CONTAINER" 6379/tcp | head -1 | awk -F: '{print $NF}')
-          fi
-          if [ -z "$REDIS_PORT" ]; then
-            echo "::error::Could not resolve host port for $REDIS_CONTAINER"
-            docker port "$REDIS_CONTAINER" 6379/tcp || true
-            docker logs "$REDIS_CONTAINER" || true
-            exit 1
-          fi
-          echo "REDIS_PORT=${REDIS_PORT}" >> "$GITHUB_ENV"
-          echo "REDIS_URL=redis://127.0.0.1:${REDIS_PORT}" >> "$GITHUB_ENV"
-          echo "Redis host port: ${REDIS_PORT}"
-          for i in $(seq 1 15); do
-            if docker exec "$REDIS_CONTAINER" redis-cli ping 2>/dev/null | grep -q PONG; then
-              echo "Redis ready after ${i}s"
-              exit 0
-            fi
-            sleep 1
-          done
-          echo "::error::Redis did not become ready in 15s"
-          docker logs "$REDIS_CONTAINER" || true
-          exit 1
-      - name: Build platform
-        if: needs.detect-changes.outputs.api == 'true'
-        working-directory: workspace-server
-        run: go build -o platform-server ./cmd/server
-      - name: Start platform (background)
-        if: needs.detect-changes.outputs.api == 'true'
-        working-directory: workspace-server
-        run: |
-          # DATABASE_URL + REDIS_URL exported by the start-postgres /
-          # start-redis steps point at this run's per-run host ports.
-          ./platform-server > platform.log 2>&1 &
-          echo $! > platform.pid
-      - name: Wait for /health
-        if: needs.detect-changes.outputs.api == 'true'
-        run: |
-          for i in $(seq 1 30); do
-            if curl -sf http://127.0.0.1:8080/health > /dev/null; then
-              echo "Platform up after ${i}s"
-              exit 0
-            fi
-            sleep 1
-          done
-          echo "::error::Platform did not become healthy in 30s"
-          cat workspace-server/platform.log || true
-          exit 1
-      - name: Assert migrations applied
-        if: needs.detect-changes.outputs.api == 'true'
-        run: |
-          tables=$(docker exec "$PG_CONTAINER" psql -U dev -d molecule -tAc "SELECT count(*) FROM information_schema.tables WHERE table_schema='public' AND table_name='workspaces'")
-          if [ "$tables" != "1" ]; then
-            echo "::error::Migrations did not apply"
-            cat workspace-server/platform.log || true
-            exit 1
-          fi
-          echo "Migrations OK"
-      - name: Run E2E API tests
-        if: needs.detect-changes.outputs.api == 'true'
-        run: bash tests/e2e/test_api.sh
-      - name: Run notify-with-attachments E2E
-        if: needs.detect-changes.outputs.api == 'true'
-        run: bash tests/e2e/test_notify_attachments_e2e.sh
-      - name: Run priority-runtimes E2E (claude-code + hermes — skips when keys absent)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: bash tests/e2e/test_priority_runtimes_e2e.sh
-      - name: Run poll-mode + since_id cursor E2E (#2339)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: bash tests/e2e/test_poll_mode_e2e.sh
-      - name: Run poll-mode chat upload E2E (RFC #2891)
-        if: needs.detect-changes.outputs.api == 'true'
-        run: bash tests/e2e/test_poll_mode_chat_upload_e2e.sh
-      - name: Dump platform log on failure
-        if: failure() && needs.detect-changes.outputs.api == 'true'
-        run: cat workspace-server/platform.log || true
-      - name: Stop platform
-        if: always() && needs.detect-changes.outputs.api == 'true'
-        run: |
-          if [ -f workspace-server/platform.pid ]; then
-            kill "$(cat workspace-server/platform.pid)" 2>/dev/null || true
-          fi
-      - name: Stop service containers
-        # always() so containers don't leak when test steps fail. The
-        # cleanup is best-effort: if the container is already gone
-        # (e.g. concurrent rerun race), don't fail the job.
-        if: always() && needs.detect-changes.outputs.api == 'true'
-        run: |
-          docker rm -f "$PG_CONTAINER" 2>/dev/null || true
-          docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true
@@ -1,216 +0,0 @@
-name: E2E Staging Canvas (Playwright)
-
-# Playwright test suite that provisions a fresh staging org per run and
-# verifies every workspace-panel tab renders without crashing. Complements
-# e2e-staging-saas.yml (which tests the API shape) by exercising the
-# actual browser + canvas bundle against live staging.
-#
-# Triggers: push to main/staging or PR touching canvas sources + this workflow,
-# manual dispatch, and weekly cron to catch browser/runtime drift even
-# when canvas is quiet.
-# Added staging to push/pull_request branches so the auto-promote gate
-# check (--event push --branch staging) can see a completed run for this
-# workflow — mirrors what PR #1891 does for e2e-api.yml.
-
-on:
-  # Trigger model (revised 2026-04-29):
-  #
-  # Always fires on push/pull_request; real work is gated per-step on
-  # `needs.detect-changes.outputs.canvas`. When canvas/ paths haven't
-  # changed, the no-op step alone runs and emits SUCCESS for the
-  # `Canvas tabs E2E` check, satisfying branch protection without
-  # spending CI cycles. See e2e-api.yml for the rationale on why this
-  # is a single job rather than two-jobs-sharing-name.
-  push:
-    branches: [main]
-  pull_request:
-    branches: [main]
-  workflow_dispatch:
-  schedule:
-    # Weekly on Sunday 08:00 UTC — catches Chrome / Playwright / Next.js
-    # release-note-shaped regressions that don't ride in with a PR.
-    - cron: '0 8 * * 0'
-
-concurrency:
-  # Per-SHA grouping (changed 2026-04-28 from a single global group). The
-  # global group made auto-promote-staging brittle: when a staging push
-  # queued behind an in-flight run and a third entrant (a PR run, a
-  # follow-on push) entered the group, the staging push got cancelled —
-  # leaving auto-promote-staging looking at `completed/cancelled` for a
-  # required gate and refusing to advance main. Observed 2026-04-28
-  # 23:51-23:53 on staging tip 3f99fede.
-  #
-  # The original intent of the global group was to throttle parallel
-  # E2E provisions (each spins a fresh EC2). At our scale that throttle
-  # isn't worth the correctness cost — fresh-org-per-run isolates the
-  # state, and the cost of two parallel runs (~$0.001/min × 10min × 2)
-  # is rounding error vs. the cost of a stuck pipeline.
-  #
-  # Per-SHA still dedupes accidental double-triggers for the SAME SHA.
-  # It does NOT cancel obsolete-PR-version runs on force-push; that
-  # wasted CI is acceptable given the alternative is losing staging-tip
-  # data that auto-promote-staging needs.
-  group: e2e-staging-canvas-${{ github.event.pull_request.head.sha || github.sha }}
-  cancel-in-progress: false
-
-jobs:
-  detect-changes:
-    runs-on: ubuntu-latest
-    outputs:
-      canvas: ${{ steps.decide.outputs.canvas }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
-        id: filter
-        with:
-          filters: |
-            canvas:
-              - 'canvas/**'
-              - '.github/workflows/e2e-staging-canvas.yml'
-      - id: decide
-        # Always run real tests for manual dispatch and the weekly cron —
-        # both exist precisely to exercise the suite, regardless of diff.
-        run: |
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ] || [ "${{ github.event_name }}" = "schedule" ]; then
-            echo "canvas=true" >> "$GITHUB_OUTPUT"
-          else
-            echo "canvas=${{ steps.filter.outputs.canvas }}" >> "$GITHUB_OUTPUT"
-          fi
-
-  # ONE job (no job-level `if:`) that always runs and reports under the
-  # required-check name `Canvas tabs E2E`. Real work is gated per-step on
-  # `needs.detect-changes.outputs.canvas`. See e2e-api.yml for the full
-  # rationale — same path-filter check-name parity issue blocked PR #2264
-  # (staging→main) on 2026-04-29 because branch protection treats matching-
-  # name check runs as a SET, and any SKIPPED member fails the eval.
-  playwright:
-    needs: detect-changes
-    name: Canvas tabs E2E
-    runs-on: ubuntu-latest
-    timeout-minutes: 40
-
-    env:
-      CANVAS_E2E_STAGING: '1'
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-
-    defaults:
-      run:
-        working-directory: canvas
-
-    steps:
-      - name: No-op pass (paths filter excluded this commit)
-        if: needs.detect-changes.outputs.canvas != 'true'
-        working-directory: .
-        run: |
-          echo "No canvas / workflow changes — E2E Staging Canvas gate satisfied without running tests."
-          echo "::notice::E2E Staging Canvas no-op pass (paths filter excluded this commit)."
-
-      - if: needs.detect-changes.outputs.canvas == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify admin token present
-        if: needs.detect-changes.outputs.canvas == 'true'
-        run: |
-          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
-            echo "::error::Missing MOLECULE_STAGING_ADMIN_TOKEN"
-            exit 2
-          fi
-
-      - name: Set up Node
-        if: needs.detect-changes.outputs.canvas == 'true'
-        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
-        with:
-          node-version: '20'
-          cache: 'npm'
-          cache-dependency-path: canvas/package-lock.json
-
-      - name: Install canvas deps
-        if: needs.detect-changes.outputs.canvas == 'true'
-        run: npm ci
-
-      - name: Install Playwright browsers
-        if: needs.detect-changes.outputs.canvas == 'true'
-        timeout-minutes: 10
-        run: npx playwright install --with-deps chromium
-
-      - name: Run staging canvas E2E
-        if: needs.detect-changes.outputs.canvas == 'true'
-        run: npx playwright test --config=playwright.staging.config.ts
-
-      - name: Upload Playwright report on failure
-        if: failure() && needs.detect-changes.outputs.canvas == 'true'
-        # Pinned to v3 for Gitea act_runner v0.6 compatibility — v4+ uses
-        # the GHES 3.10+ artifact protocol that Gitea 1.22.x does NOT
-        # implement (see ci.yml upload step for the canonical error
-        # cite). Drop this pin when Gitea ships the v4 protocol.
-        uses: actions/upload-artifact@c6a366c94c3e0affe28c06c8df20a878f24da3cf # v3.2.2
-        with:
-          name: playwright-report-staging
-          path: canvas/playwright-report-staging/
-          retention-days: 14
-
-      - name: Upload screenshots on failure
-        if: failure() && needs.detect-changes.outputs.canvas == 'true'
-        # Pinned to v3 for Gitea act_runner v0.6 compatibility (see above).
-        uses: actions/upload-artifact@c6a366c94c3e0affe28c06c8df20a878f24da3cf # v3.2.2
-        with:
-          name: playwright-screenshots
-          path: canvas/test-results/
-          retention-days: 14
-
-      # Safety-net teardown — fires only when Playwright's globalTeardown
-      # didn't (worker crash, runner cancel). Reads the slug from
-      # canvas/.playwright-staging-state.json (written by staging-setup
-      # as its first action, before any CP call) and deletes only that
-      # slug.
-      #
-      # Earlier versions of this step pattern-swept `e2e-canvas-<today>-*`
-      # orgs to compensate for setup-crash-before-state-file-write. That
-      # over-aggressive cleanup raced concurrent canvas-E2E runs and
-      # poisoned each other's tenants — observed 2026-04-30 when three
-      # real-test runs killed each other mid-test, surfacing as
-      # `getaddrinfo ENOTFOUND` once CP had cleaned up the just-deleted
-      # DNS record. Pattern-sweep removed; setup now writes the state
-      # file before any CP work, so the slug is always recoverable.
-      - name: Teardown safety net
-        if: always() && needs.detect-changes.outputs.canvas == 'true'
-        env:
-          ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-        run: |
-          set +e
-          STATE_FILE=".playwright-staging-state.json"
-          if [ ! -f "$STATE_FILE" ]; then
-            echo "::notice::No state file at canvas/$STATE_FILE — Playwright globalTeardown handled it (or setup never ran)."
-            exit 0
-          fi
-          slug=$(python3 -c "import json; print(json.load(open('$STATE_FILE')).get('slug',''))")
-          if [ -z "$slug" ]; then
-            echo "::warning::State file present but slug missing; nothing to clean up."
-            exit 0
-          fi
-          echo "Deleting orphan tenant: $slug"
-          # Verify HTTP 2xx instead of `>/dev/null || true` swallowing
-          # failures. A 5xx or timeout previously looked identical to
-          # success, leaving the tenant alive for up to ~45 min until
-          # sweep-stale-e2e-orgs caught it. Surface failures as
-          # workflow warnings naming the slug. Don't `exit 1` — a single
-          # cleanup miss shouldn't fail-flag the canvas test when the
-          # actual smoke check passed; the sweeper is the safety net.
-          # See molecule-controlplane#420.
-          # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-          # pollution of the captured status (lint-curl-status-capture.yml).
-          set +e
-          curl -sS -o /tmp/canvas-cleanup.out -w "%{http_code}" \
-            -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" \
-            -H "Content-Type: application/json" \
-            -d "{\"confirm\":\"$slug\"}" >/tmp/canvas-cleanup.code
-          set -e
-          code=$(cat /tmp/canvas-cleanup.code 2>/dev/null || echo "000")
-          if [ "$code" = "200" ] || [ "$code" = "204" ]; then
-            echo "[teardown] deleted $slug (HTTP $code)"
-          else
-            echo "::warning::canvas teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within ~45 min. Body: $(head -c 300 /tmp/canvas-cleanup.out 2>/dev/null)"
-          fi
-          exit 0
@@ -1,184 +0,0 @@
-name: E2E Staging External Runtime
-
-# Regression for the four/five workspaces.status=awaiting_agent transitions
-# that silently failed in production for five days before migration 046
-# extended the workspace_status enum (see
-# workspace-server/migrations/046_workspace_status_awaiting_agent.up.sql).
-#
-# Why this is its own workflow (not folded into e2e-staging-saas.yml):
-#   - The full-saas harness defaults to runtime=hermes, never exercises
-#     external-runtime. Adding an `external` parameter to that script
-#     would force every push to staging through both lifecycles in
-#     series, doubling the EC2 cold-start budget.
-#   - The external lifecycle has unique timing (REMOTE_LIVENESS_STALE_AFTER
-#     window, 90s default + sweep interval), which we wait through
-#     deliberately. Folding it into hermes would make the long path
-#     even longer.
-#   - It can run in parallel with the hermes E2E since both create
-#     fresh tenant orgs with distinct slug prefixes (`e2e-ext-...` vs
-#     `e2e-...`).
-#
-# Triggers:
-#   - Push to staging when any source affecting external runtime,
-#     hibernation, or the migration set changes.
-#   - PR review for the same set.
-#   - Manual workflow_dispatch.
-#   - Daily cron at 07:30 UTC (catches drift on quiet days; staggered
-#     30 min after e2e-staging-saas.yml's 07:00 UTC cron).
-#
-# Concurrency: serialized so two staging pushes don't fight for the
-# same EC2 quota window. cancel-in-progress=false so a half-rolled
-# tenant always finishes its teardown.
-
-on:
-  push:
-    branches: [main]
-    paths:
-      - 'workspace-server/internal/handlers/workspace.go'
-      - 'workspace-server/internal/handlers/registry.go'
-      - 'workspace-server/internal/handlers/workspace_restart.go'
-      - 'workspace-server/internal/registry/healthsweep.go'
-      - 'workspace-server/internal/registry/liveness.go'
-      - 'workspace-server/migrations/**'
-      - 'workspace-server/internal/db/workspace_status_enum_drift_test.go'
-      - 'tests/e2e/test_staging_external_runtime.sh'
-      - '.github/workflows/e2e-staging-external.yml'
-  pull_request:
-    branches: [main]
-    paths:
-      - 'workspace-server/internal/handlers/workspace.go'
-      - 'workspace-server/internal/handlers/registry.go'
-      - 'workspace-server/internal/handlers/workspace_restart.go'
-      - 'workspace-server/internal/registry/healthsweep.go'
-      - 'workspace-server/internal/registry/liveness.go'
-      - 'workspace-server/migrations/**'
-      - 'workspace-server/internal/db/workspace_status_enum_drift_test.go'
-      - 'tests/e2e/test_staging_external_runtime.sh'
-      - '.github/workflows/e2e-staging-external.yml'
-  workflow_dispatch:
-    inputs:
-      keep_org:
-        description: "Skip teardown for debugging (only via manual dispatch)"
-        required: false
-        type: boolean
-        default: false
-      stale_wait_secs:
-        description: "Seconds to wait for the heartbeat-staleness sweep (default 180 = 90s window + 90s buffer)"
-        required: false
-        default: "180"
-  schedule:
-    - cron: '30 7 * * *'
-
-concurrency:
-  group: e2e-staging-external
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  e2e-staging-external:
-    name: E2E Staging External Runtime
-    runs-on: ubuntu-latest
-    timeout-minutes: 25
-
-    env:
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-      E2E_RUN_ID: "${{ github.run_id }}-${{ github.run_attempt }}"
-      E2E_KEEP_ORG: ${{ github.event.inputs.keep_org && '1' || '0' }}
-      E2E_STALE_WAIT_SECS: ${{ github.event.inputs.stale_wait_secs || '180' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify admin token present
-        run: |
-          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
-            # Schedule + push triggers must hard-fail when the token is
-            # missing — silent skip would mask infra rot. Manual dispatch
-            # gets the same hard-fail; an operator running this on a fork
-            # without secrets configured needs to know up-front.
-            echo "::error::MOLECULE_STAGING_ADMIN_TOKEN secret not set (Railway staging CP_ADMIN_API_TOKEN)"
-            exit 2
-          fi
-          echo "Admin token present ✓"
-
-      - name: CP staging health preflight
-        run: |
-          code=$(curl -sS -o /dev/null -w "%{http_code}" --max-time 10 "$MOLECULE_CP_URL/health")
-          if [ "$code" != "200" ]; then
-            echo "::error::Staging CP unhealthy (got HTTP $code). Skipping — not a workspace bug."
-            exit 1
-          fi
-          echo "Staging CP healthy ✓"
-
-      - name: Run external-runtime E2E
-        id: e2e
-        run: bash tests/e2e/test_staging_external_runtime.sh
-
-      # Mirror the e2e-staging-saas.yml safety net: if the runner is
-      # cancelled (e.g. concurrent staging push), the test script's
-      # EXIT trap may not fire, so we sweep e2e-ext-* slugs scoped to
-      # *this* run id.
-      - name: Teardown safety net (runs on cancel/failure)
-        if: always()
-        env:
-          ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-        run: |
-          set +e
-          orgs=$(curl -sS "$MOLECULE_CP_URL/cp/admin/orgs" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" 2>/dev/null \
-            | python3 -c "
-          import json, sys, os, datetime
-          run_id = os.environ.get('GITHUB_RUN_ID', '')
-          d = json.load(sys.stdin)
-          # Scope STRICTLY to this run id (e2e-ext-YYYYMMDD-<runid>-...)
-          # so concurrent runs and unrelated dev probes are not touched.
-          # Sweep today AND yesterday so a midnight-crossing run still
-          # cleans up its own slug.
-          today = datetime.date.today()
-          yesterday = today - datetime.timedelta(days=1)
-          dates = (today.strftime('%Y%m%d'), yesterday.strftime('%Y%m%d'))
-          if not run_id:
-              # Without a run id we cannot scope safely; bail rather
-              # than risk deleting unrelated tenants.
-              sys.exit(0)
-          prefixes = tuple(f'e2e-ext-{d}-{run_id}-' for d in dates)
-          for o in d.get('orgs', []):
-              s = o.get('slug', '')
-              if s.startswith(prefixes) and o.get('status') != 'purged':
-                  print(s)
-          " 2>/dev/null)
-          if [ -n "$orgs" ]; then
-            echo "Safety-net sweep: deleting leftover orgs:"
-            echo "$orgs"
-            # Per-slug verified DELETE — see molecule-controlplane#420.
-            # `>/dev/null 2>&1` previously hid every failure; surface
-            # non-2xx as workflow warnings so the run page names what
-            # leaked. Sweeper catches the rest within ~45 min.
-            leaks=()
-            for slug in $orgs; do
-              # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-              # pollution of the captured status (lint-curl-status-capture.yml).
-              set +e
-              curl -sS -o /tmp/external-cleanup.out -w "%{http_code}" \
-                -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-                -H "Authorization: Bearer $ADMIN_TOKEN" \
-                -H "Content-Type: application/json" \
-                -d "{\"confirm\":\"$slug\"}" >/tmp/external-cleanup.code
-              set -e
-              code=$(cat /tmp/external-cleanup.code 2>/dev/null || echo "000")
-              if [ "$code" = "200" ] || [ "$code" = "204" ]; then
-                echo "[teardown] deleted $slug (HTTP $code)"
-              else
-                echo "::warning::external teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within ~45 min. Body: $(head -c 300 /tmp/external-cleanup.out 2>/dev/null)"
-                leaks+=("$slug")
-              fi
-            done
-            if [ ${#leaks[@]} -gt 0 ]; then
-              echo "::warning::external teardown left ${#leaks[@]} leak(s): ${leaks[*]}"
-            fi
-          else
-            echo "Safety-net sweep: no leftover orgs to clean."
-          fi
@@ -1,246 +0,0 @@
-name: E2E Staging SaaS (full lifecycle)
-
-# Dedicated workflow that provisions a fresh staging org per run, exercises
-# the full workspace lifecycle (register → heartbeat → A2A → delegation →
-# HMA memory → activity → peers), then tears down and asserts leak-free.
-#
-# Why a separate workflow (not folded into ci.yml):
-#   - The run takes ~25-35 min (EC2 boot + cloudflared DNS + provision sweeps +
-#     agent bootstrap), way too slow for every PR.
-#   - Needs its own concurrency group so two pushes don't fight over the
-#     same staging org slug prefix.
-#   - Has its own required secrets (session cookie, admin token) that most
-#     PRs don't need to read.
-#
-# Triggers:
-#   - Push to main (regression guard)
-#   - workflow_dispatch (manual re-run from UI)
-#   - Nightly cron (catches drift even when no pushes land)
-#   - Changes to any provisioning-critical file under PR review (opt-in
-#     via the same paths watcher that e2e-api.yml uses)
-
-on:
-  # Trunk-based (Phase 3 of internal#81): main is the only branch.
-  # Previously this fired on staging push too because staging was a
-  # superset of main and ran the gate ahead of auto-promote; with no
-  # staging branch, main is where E2E gates the deploy.
-  push:
-    branches: [main]
-    paths:
-      - 'workspace-server/internal/handlers/registry.go'
-      - 'workspace-server/internal/handlers/workspace_provision.go'
-      - 'workspace-server/internal/handlers/a2a_proxy.go'
-      - 'workspace-server/internal/middleware/**'
-      - 'workspace-server/internal/provisioner/**'
-      - 'tests/e2e/test_staging_full_saas.sh'
-      - '.github/workflows/e2e-staging-saas.yml'
-  pull_request:
-    branches: [main]
-    paths:
-      - 'workspace-server/internal/handlers/registry.go'
-      - 'workspace-server/internal/handlers/workspace_provision.go'
-      - 'workspace-server/internal/handlers/a2a_proxy.go'
-      - 'workspace-server/internal/middleware/**'
-      - 'workspace-server/internal/provisioner/**'
-      - 'tests/e2e/test_staging_full_saas.sh'
-      - '.github/workflows/e2e-staging-saas.yml'
-  workflow_dispatch:
-    inputs:
-      runtime:
-        description: "Runtime to test (claude-code [default, MiniMax] | hermes [OpenAI] | langgraph [OpenAI])"
-        required: false
-        default: "claude-code"
-      keep_org:
-        description: "Skip teardown for debugging (only use via manual dispatch!)"
-        required: false
-        type: boolean
-        default: false
-  schedule:
-    # 07:00 UTC every day — catches AMI drift, WorkOS cert rotation,
-    # Cloudflare API regressions, etc. even on quiet days.
-    - cron: '0 7 * * *'
-
-# Serialize: staging has a finite per-hour org creation quota. Two pushes
-# landing in quick succession should queue, not race. `cancel-in-progress:
-# false` mirrors e2e-api.yml — GitHub would otherwise cancel the running
-# teardown step and leave orphan EC2s.
-concurrency:
-  group: e2e-staging-saas
-  cancel-in-progress: false
-
-jobs:
-  e2e-staging-saas:
-    name: E2E Staging SaaS
-    runs-on: ubuntu-latest
-    timeout-minutes: 45
-    permissions:
-      contents: read
-
-    env:
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      # Single admin-bearer secret drives provision + tenant-token
-      # retrieval + teardown. Configure in
-      # Settings → Secrets and variables → Actions → Repository secrets.
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-      # MiniMax is the PRIMARY LLM auth path post-2026-05-04. Switched
-      # from hermes+OpenAI default after #2578 (the staging OpenAI key
-      # account went over quota and stayed dead for 36+ hours, taking
-      # the full-lifecycle E2E red on every provisioning-critical push).
-      # claude-code template's `minimax` provider routes
-      # ANTHROPIC_BASE_URL to api.minimax.io/anthropic and reads
-      # MINIMAX_API_KEY at boot — separate billing account so an
-      # OpenAI quota collapse no longer wedges the gate. Mirrors the
-      # canary-staging.yml + continuous-synth-e2e.yml migrations.
-      E2E_MINIMAX_API_KEY: ${{ secrets.MOLECULE_STAGING_MINIMAX_API_KEY }}
-      # Direct-Anthropic alternative for operators who don't want to
-      # set up a MiniMax account (priority below MiniMax — first
-      # non-empty wins in test_staging_full_saas.sh's secrets-injection
-      # block). See #2578 PR comment for the rationale.
-      E2E_ANTHROPIC_API_KEY: ${{ secrets.MOLECULE_STAGING_ANTHROPIC_API_KEY }}
-      # OpenAI fallback — kept wired so an operator-dispatched run with
-      # E2E_RUNTIME=hermes or =langgraph via workflow_dispatch can still
-      # exercise the OpenAI path.
-      E2E_OPENAI_API_KEY: ${{ secrets.MOLECULE_STAGING_OPENAI_KEY }}
-      E2E_RUNTIME: ${{ github.event.inputs.runtime || 'claude-code' }}
-      # Pin the model when running on the default claude-code path —
-      # the per-runtime default ("sonnet") routes to direct Anthropic
-      # and defeats the cost saving. Operators can override via the
-      # workflow_dispatch flow (no input wired here yet — runtime
-      # override is enough for ad-hoc).
-      E2E_MODEL_SLUG: ${{ github.event.inputs.runtime == 'hermes' && 'openai/gpt-4o' || github.event.inputs.runtime == 'langgraph' && 'openai:gpt-4o' || 'MiniMax-M2.7-highspeed' }}
-      E2E_RUN_ID: "${{ github.run_id }}-${{ github.run_attempt }}"
-      E2E_KEEP_ORG: ${{ github.event.inputs.keep_org && '1' || '0' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify admin token present
-        run: |
-          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
-            echo "::error::MOLECULE_STAGING_ADMIN_TOKEN secret not set (Railway staging CP_ADMIN_API_TOKEN)"
-            exit 2
-          fi
-          echo "Admin token present ✓"
-
-      - name: Verify LLM key present
-        run: |
-          # Per-runtime key check — claude-code uses MiniMax; hermes /
-          # langgraph (operator-dispatched only) use OpenAI. Hard-fail
-          # rather than soft-skip per #2578's lesson — empty key
-          # silently falls through to the wrong SECRETS_JSON branch and
-          # produces a confusing auth error 5 min later instead of the
-          # clean "secret missing" message at the top.
-          case "${E2E_RUNTIME}" in
-            claude-code)
-              # Either MiniMax OR direct-Anthropic works — first
-              # non-empty wins in the test script's secrets-injection
-              # priority chain.
-              if [ -n "${E2E_MINIMAX_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY"
-                required_secret_value="${E2E_MINIMAX_API_KEY}"
-              elif [ -n "${E2E_ANTHROPIC_API_KEY:-}" ]; then
-                required_secret_name="MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value="${E2E_ANTHROPIC_API_KEY}"
-              else
-                required_secret_name="MOLECULE_STAGING_MINIMAX_API_KEY or MOLECULE_STAGING_ANTHROPIC_API_KEY"
-                required_secret_value=""
-              fi
-              ;;
-            langgraph|hermes)
-              required_secret_name="MOLECULE_STAGING_OPENAI_KEY"
-              required_secret_value="${E2E_OPENAI_API_KEY:-}"
-              ;;
-            *)
-              echo "::warning::Unknown E2E_RUNTIME='${E2E_RUNTIME}' — skipping LLM-key check"
-              required_secret_name=""
-              required_secret_value="present"
-              ;;
-          esac
-          if [ -n "$required_secret_name" ] && [ -z "$required_secret_value" ]; then
-            echo "::error::${required_secret_name} secret not set for runtime=${E2E_RUNTIME} — workspaces will fail at boot with 'No provider API key found'"
-            exit 2
-          fi
-          echo "LLM key present ✓ (runtime=${E2E_RUNTIME}, key=${required_secret_name}, len=${#required_secret_value})"
-
-      - name: CP staging health preflight
-        run: |
-          code=$(curl -sS -o /dev/null -w "%{http_code}" --max-time 10 "$MOLECULE_CP_URL/health")
-          if [ "$code" != "200" ]; then
-            echo "::error::Staging CP unhealthy (got HTTP $code). Skipping — not a workspace bug."
-            exit 1
-          fi
-          echo "Staging CP healthy ✓"
-
-      - name: Run full-lifecycle E2E
-        id: e2e
-        run: bash tests/e2e/test_staging_full_saas.sh
-
-      # Belt-and-braces teardown: the test script itself installs a trap
-      # for EXIT/INT/TERM, but if the GH runner itself is cancelled (e.g.
-      # someone pushes a new commit and workflow concurrency is set to
-      # cancel), the trap may not fire. This `always()` step runs even on
-      # cancellation and attempts the delete a second time. The admin
-      # DELETE endpoint is idempotent so double-invoking is safe.
-      - name: Teardown safety net (runs on cancel/failure)
-        if: always()
-        env:
-          ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-        run: |
-          # Best-effort: find any e2e-YYYYMMDD-* orgs matching this run and
-          # nuke them. Catches the case where the script died before
-          # exporting its slug.
-          set +e
-          orgs=$(curl -sS "$MOLECULE_CP_URL/cp/admin/orgs" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" 2>/dev/null \
-            | python3 -c "
-          import json, sys, os, datetime
-          run_id = os.environ.get('GITHUB_RUN_ID', '')
-          d = json.load(sys.stdin)
-          # ONLY sweep slugs from *this* CI run. Previously the filter was
-          # f'e2e-{today}-' which stomped on parallel CI runs AND any manual
-          # E2E probes a dev was running against staging (incident 2026-04-21
-          # 15:02Z: this workflow's safety net deleted an unrelated manual
-          # run's tenant 1s after it hit 'running').
-          # Sweep both today AND yesterday's UTC dates so a run that crosses
-          # midnight still matches its own slug — see the 2026-04-26→27
-          # canvas-safety-net incident for the same bug class.
-          today = datetime.date.today()
-          yesterday = today - datetime.timedelta(days=1)
-          dates = (today.strftime('%Y%m%d'), yesterday.strftime('%Y%m%d'))
-          if run_id:
-              prefixes = tuple(f'e2e-{d}-{run_id}-' for d in dates)
-          else:
-              prefixes = tuple(f'e2e-{d}-' for d in dates)
-          candidates = [o['slug'] for o in d.get('orgs', [])
-                        if any(o.get('slug','').startswith(p) for p in prefixes)
-                        and o.get('instance_status') not in ('purged',)]
-          print('\n'.join(candidates))
-          " 2>/dev/null)
-          # Per-slug verified DELETE (was `>/dev/null || true` — see
-          # molecule-controlplane#420). Surface non-2xx as a workflow
-          # warning naming the leaked slug; don't exit 1 (sweeper is
-          # the safety net within ~45 min).
-          leaks=()
-          for slug in $orgs; do
-            echo "Safety-net teardown: $slug"
-            # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-            # pollution of the captured status (lint-curl-status-capture.yml).
-            set +e
-            curl -sS -o /tmp/saas-cleanup.out -w "%{http_code}" \
-              -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-              -H "Authorization: Bearer $ADMIN_TOKEN" \
-              -H "Content-Type: application/json" \
-              -d "{\"confirm\":\"$slug\"}" >/tmp/saas-cleanup.code
-            set -e
-            code=$(cat /tmp/saas-cleanup.code 2>/dev/null || echo "000")
-            if [ "$code" = "200" ] || [ "$code" = "204" ]; then
-              echo "[teardown] deleted $slug (HTTP $code)"
-            else
-              echo "::warning::saas teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within ~45 min. Body: $(head -c 300 /tmp/saas-cleanup.out 2>/dev/null)"
-              leaks+=("$slug")
-            fi
-          done
-          if [ ${#leaks[@]} -gt 0 ]; then
-            echo "::warning::saas teardown left ${#leaks[@]} leak(s): ${leaks[*]}"
-          fi
-          exit 0
@@ -1,171 +0,0 @@
-name: E2E Staging Sanity (leak-detection self-check)
-
-# Periodic assertion that the teardown safety nets in e2e-staging-saas
-# and canary-staging actually work. Runs the E2E harness with
-# E2E_INTENTIONAL_FAILURE=1, which poisons the tenant admin token after
-# the org is provisioned. The workspace-provision step then fails, the
-# script exits non-zero, and the EXIT trap + workflow always()-step
-# must still tear down cleanly.
-#
-# A green run means:
-#   - The script exited non-zero (intentional failure caught)
-#   - The trap fired teardown
-#   - The leak-detection poll found zero orphan orgs
-#
-# A red run means the teardown path itself is broken — act on this the
-# same way you'd act on a canary failure (the whole E2E safety net is
-# compromised until it's fixed).
-#
-# Cadence: once a week, Monday 06:00 UTC. Drift-slow, not per-PR — the
-# teardown path rarely changes, and a weekly heartbeat is enough to
-# catch silent regressions in cleanup code paths.
-
-on:
-  schedule:
-    - cron: '0 6 * * 1'
-  workflow_dispatch:
-
-concurrency:
-  # Shares the group with canary + full so they don't collide on
-  # staging org-create quota.
-  group: e2e-staging-sanity
-  cancel-in-progress: false
-
-permissions:
-  issues: write
-  contents: read
-
-jobs:
-  sanity:
-    name: Intentional-failure teardown sanity
-    runs-on: ubuntu-latest
-    timeout-minutes: 20
-
-    env:
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      MOLECULE_ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-      E2E_MODE: canary            # lean lifecycle; we only need the org to exist
-      E2E_RUNTIME: hermes
-      E2E_RUN_ID: "sanity-${{ github.run_id }}"
-      E2E_INTENTIONAL_FAILURE: "1"
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify admin token present
-        run: |
-          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
-            echo "::error::MOLECULE_STAGING_ADMIN_TOKEN not set"
-            exit 2
-          fi
-
-      # Inverted assertion: the run MUST fail. If it passes, the
-      # E2E_INTENTIONAL_FAILURE path is broken (token not being
-      # poisoned correctly, or the harness silently recovered).
-      - name: Run harness — expecting exit !=0
-        id: harness
-        run: |
-          set +e
-          bash tests/e2e/test_staging_full_saas.sh
-          rc=$?
-          echo "harness_rc=$rc" >> "$GITHUB_OUTPUT"
-          # The only acceptable outcomes:
-          #   1 — harness failed mid-run, teardown ran, leak-check passed
-          #   (exit 4 means teardown left a leak — that's the real bug
-          #    this sanity check exists to catch)
-          if [ "$rc" = "1" ]; then
-            echo "✓ Harness failed as expected (rc=1); teardown trap ran, leak-check passed"
-            exit 0
-          elif [ "$rc" = "0" ]; then
-            echo "::error::Harness succeeded under E2E_INTENTIONAL_FAILURE=1 — the poisoning path is broken"
-            exit 1
-          elif [ "$rc" = "4" ]; then
-            echo "::error::LEAK DETECTED (rc=4) — teardown failed to clean up the org. Safety net broken."
-            exit 4
-          else
-            echo "::error::Unexpected rc=$rc — neither clean-failure nor leak. Investigate harness."
-            exit 1
-          fi
-
-      - name: Open issue if safety net is broken
-        if: failure()
-        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
-        with:
-          script: |
-            const title = "🚨 E2E teardown safety net broken";
-            const runURL = `https://github.com/${context.repo.owner}/${context.repo.repo}/actions/runs/${context.runId}`;
-            const body =
-              `The weekly sanity run (E2E_INTENTIONAL_FAILURE=1) did not exit ` +
-              `as expected. This means one of:\n` +
-              `  - poisoning didn't actually cause failure (test harness regression), OR\n` +
-              `  - teardown left an orphan org (leak detection caught a real bug)\n\n` +
-              `Run: ${runURL}\n\n` +
-              `This is higher priority than a canary failure — the whole ` +
-              `E2E safety net can't be trusted until this is resolved.`;
-
-            const { data: existing } = await github.rest.issues.listForRepo({
-              owner: context.repo.owner, repo: context.repo.repo,
-              state: 'open', labels: 'e2e-safety-net',
-            });
-            const match = existing.find(i => i.title === title);
-            if (match) {
-              await github.rest.issues.createComment({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: match.number,
-                body: `Still broken. ${runURL}`,
-              });
-            } else {
-              await github.rest.issues.create({
-                owner: context.repo.owner, repo: context.repo.repo,
-                title, body,
-                labels: ['e2e-safety-net', 'bug', 'priority-high'],
-              });
-            }
-
-      # Belt-and-braces: if teardown left anything behind, nuke it here
-      # so we don't bleed staging quota. Different label from the
-      # always()-steps in the other workflows so sanity-only orgs get
-      # cleaned up by sanity runs.
-      - name: Teardown safety net
-        if: always()
-        env:
-          ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-        run: |
-          set +e
-          orgs=$(curl -sS "$MOLECULE_CP_URL/cp/admin/orgs" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" 2>/dev/null \
-            | python3 -c "
-          import json, sys
-          d = json.load(sys.stdin)
-          today = __import__('datetime').date.today().strftime('%Y%m%d')
-          candidates = [o['slug'] for o in d.get('orgs', [])
-                        if o.get('slug','').startswith(f'e2e-canary-{today}-sanity-')
-                        and o.get('status') not in ('purged',)]
-          print('\n'.join(candidates))
-          " 2>/dev/null)
-          # Per-slug verified DELETE — see molecule-controlplane#420.
-          # Failures surface as workflow warnings; the sweeper is the
-          # safety net within ~45 min.
-          leaks=()
-          for slug in $orgs; do
-            # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-            # pollution of the captured status (lint-curl-status-capture.yml).
-            set +e
-            curl -sS -o /tmp/sanity-cleanup.out -w "%{http_code}" \
-              -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-              -H "Authorization: Bearer $ADMIN_TOKEN" \
-              -H "Content-Type: application/json" \
-              -d "{\"confirm\":\"$slug\"}" >/tmp/sanity-cleanup.code
-            set -e
-            code=$(cat /tmp/sanity-cleanup.code 2>/dev/null || echo "000")
-            if [ "$code" = "200" ] || [ "$code" = "204" ]; then
-              echo "[teardown] deleted $slug (HTTP $code)"
-            else
-              echo "::warning::sanity teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within ~45 min. Body: $(head -c 300 /tmp/sanity-cleanup.out 2>/dev/null)"
-              leaks+=("$slug")
-            fi
-          done
-          if [ ${#leaks[@]} -gt 0 ]; then
-            echo "::warning::sanity teardown left ${#leaks[@]} leak(s): ${leaks[*]}"
-          fi
-          exit 0
@@ -1,252 +0,0 @@
-name: Handlers Postgres Integration
-
-# Real-Postgres integration tests for workspace-server/internal/handlers/.
-# Triggered on every PR/push that touches the handlers package.
-#
-# Why this workflow exists
-# ------------------------
-# Strict-sqlmock unit tests pin which SQL statements fire — they're fast
-# and let us iterate without a DB. But sqlmock CANNOT detect bugs that
-# depend on the row state AFTER the SQL runs. The result_preview-lost
-# bug shipped to staging in PR #2854 because every unit test was
-# satisfied with "an UPDATE statement fired" — none verified the row's
-# preview field actually landed. The local-postgres E2E that retrofit
-# self-review caught it took 2 minutes to set up and would have caught
-# the bug at PR-time.
-#
-# Why this workflow does NOT use `services: postgres:` (Class B fix)
-# ------------------------------------------------------------------
-# Our act_runner config has `container.network: host` (operator host
-# /opt/molecule/runners/config.yaml), which act_runner applies to BOTH
-# the job container AND every service container. With host-net, two
-# concurrent runs of this workflow both try to bind 0.0.0.0:5432 — the
-# second postgres FATALs with `could not create any TCP/IP sockets:
-# Address in use`, and Docker auto-removes it (act_runner sets
-# AutoRemove:true on service containers). By the time the migrations
-# step runs `psql`, the postgres container is gone, hence
-# `Connection refused` then `failed to remove container: No such
-# container` at cleanup time.
-#
-# Per-job `container.network` override is silently ignored by
-# act_runner — `--network and --net in the options will be ignored.`
-# appears in the runner log. Documented constraint.
-#
-# So we sidestep `services:` entirely. The job container still uses
-# host-net (inherited from runner config; required for cache server
-# discovery on the bridge IP 172.18.0.17:42631). We launch a sibling
-# postgres on the existing `molecule-core-net` bridge with a
-# UNIQUE name per run — `pg-handlers-${RUN_ID}-${RUN_ATTEMPT}` — and
-# read its bridge IP via `docker inspect`. A host-net job container
-# can reach a bridge-net container directly via the bridge IP (verified
-# manually on operator host 2026-05-08).
-#
-# Trade-offs vs. the original `services:` shape:
-#   + No host-port collision; N parallel runs share the bridge cleanly
-#   + `if: always()` cleanup runs even on test-step failure
-#   - One more step in the workflow (+~3 lines)
-#   - Requires `molecule-core-net` to exist on the operator host
-#     (it does; declared in docker-compose.yml + docker-compose.infra.yml)
-#
-# Class B Hongming-owned CICD red sweep, 2026-05-08.
-#
-# Cost: ~30s job (postgres pull from cache + go build + 4 tests).
-
-on:
-  push:
-    branches: [main, staging]
-  pull_request:
-    branches: [main, staging]
-  merge_group:
-    types: [checks_requested]
-  workflow_dispatch:
-
-concurrency:
-  group: handlers-pg-integ-${{ github.event.pull_request.head.sha || github.sha }}
-  cancel-in-progress: false
-
-jobs:
-  detect-changes:
-    name: detect-changes
-    runs-on: ubuntu-latest
-    outputs:
-      handlers: ${{ steps.filter.outputs.handlers }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
-        id: filter
-        with:
-          filters: |
-            handlers:
-              - 'workspace-server/internal/handlers/**'
-              - 'workspace-server/internal/wsauth/**'
-              - 'workspace-server/migrations/**'
-              - '.github/workflows/handlers-postgres-integration.yml'
-
-  # Single-job-with-per-step-if pattern: always runs to satisfy the
-  # required-check name on branch protection; real work gates on the
-  # paths filter. See ci.yml's Platform (Go) for the same shape.
-  integration:
-    name: Handlers Postgres Integration
-    needs: detect-changes
-    runs-on: ubuntu-latest
-    env:
-      # Unique name per run so concurrent jobs don't collide on the
-      # bridge network. ${RUN_ID}-${RUN_ATTEMPT} is unique even across
-      # workflow_dispatch reruns of the same run_id.
-      PG_NAME: pg-handlers-${{ github.run_id }}-${{ github.run_attempt }}
-      # Bridge network already exists on the operator host (declared
-      # in docker-compose.yml + docker-compose.infra.yml).
-      PG_NETWORK: molecule-core-net
-    defaults:
-      run:
-        working-directory: workspace-server
-    steps:
-      - if: needs.detect-changes.outputs.handlers != 'true'
-        working-directory: .
-        run: echo "No handlers/migrations changes — skipping; this job always runs to satisfy the required-check name."
-
-      - if: needs.detect-changes.outputs.handlers == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - if: needs.detect-changes.outputs.handlers == 'true'
-        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
-        with:
-          go-version: 'stable'
-
-      - if: needs.detect-changes.outputs.handlers == 'true'
-        name: Start sibling Postgres on bridge network
-        working-directory: .
-        run: |
-          # Sanity: the bridge network must exist on the operator host.
-          # Hard-fail loud if it doesn't — easier to spot than a silent
-          # auto-create that diverges from the rest of the stack.
-          if ! docker network inspect "${PG_NETWORK}" >/dev/null 2>&1; then
-            echo "::error::Bridge network '${PG_NETWORK}' missing on operator host. Re-run docker-compose.infra.yml or check ops handbook."
-            exit 1
-          fi
-
-          # If a stale container with the same name exists (rerun on
-          # the same run_id), wipe it first.
-          docker rm -f "${PG_NAME}" >/dev/null 2>&1 || true
-
-          docker run -d \
-            --name "${PG_NAME}" \
-            --network "${PG_NETWORK}" \
-            --health-cmd "pg_isready -U postgres" \
-            --health-interval 5s \
-            --health-timeout 5s \
-            --health-retries 10 \
-            -e POSTGRES_PASSWORD=test \
-            -e POSTGRES_DB=molecule \
-            postgres:15-alpine >/dev/null
-
-          # Read back the bridge IP. Always present immediately after
-          # `docker run -d` for bridge networks.
-          PG_HOST=$(docker inspect "${PG_NAME}" \
-            --format "{{(index .NetworkSettings.Networks \"${PG_NETWORK}\").IPAddress}}")
-          if [ -z "${PG_HOST}" ]; then
-            echo "::error::Could not resolve PG_HOST for ${PG_NAME} on ${PG_NETWORK}"
-            docker logs "${PG_NAME}" || true
-            exit 1
-          fi
-          echo "PG_HOST=${PG_HOST}" >> "$GITHUB_ENV"
-          echo "INTEGRATION_DB_URL=postgres://postgres:test@${PG_HOST}:5432/molecule?sslmode=disable" >> "$GITHUB_ENV"
-          echo "Started ${PG_NAME} at ${PG_HOST}:5432"
-
-      - if: needs.detect-changes.outputs.handlers == 'true'
-        name: Apply migrations to Postgres service
-        env:
-          PGPASSWORD: test
-        run: |
-          # Wait for postgres to actually accept connections. Docker's
-          # health-cmd handles container-side readiness, but the wire
-          # to the bridge IP is best-tested with pg_isready directly.
-          for i in {1..15}; do
-            if pg_isready -h "${PG_HOST}" -p 5432 -U postgres -q; then break; fi
-            echo "waiting for postgres at ${PG_HOST}:5432..."; sleep 2
-          done
-
-          # Apply every .up.sql in lexicographic order with
-          # ON_ERROR_STOP=0 — failing migrations are SKIPPED rather than
-          # blocking the suite. This handles the current schema state
-          # where a few historical migrations (e.g. 017_memories_fts_*)
-          # depend on tables that were later renamed/dropped and so
-          # cannot replay from scratch. The migrations that DO succeed
-          # land their tables, which is sufficient for the integration
-          # tests in handlers/.
-          #
-          # Why not maintain a curated allowlist: every new migration
-          # touching a handlers/-tested table would have to update this
-          # workflow. With apply-all-or-skip, a future migration that
-          # adds a column to delegations runs automatically (its base
-          # table 049_delegations.up.sql already succeeded above it in
-          # the order). Operators only need to revisit this if the
-          # migration chain becomes legitimately replayable end-to-end.
-          #
-          # Per-migration result is logged so a failed migration that
-          # SHOULD have been replayable surfaces in the CI log instead
-          # of silently failing.
-          # Apply both *.sql (legacy, lives next to its module) and
-          # *.up.sql (newer up/down convention) in a single
-          # lexicographically-sorted pass. Excluding *.down.sql so the
-          # newest-naming-convention pairs don't undo themselves mid-run.
-          # Pre-#149-followup this loop only globbed *.up.sql, which
-          # silently skipped 001_workspaces.sql + 009_activity_logs.sql
-          # — fine while no integration test depended on those tables,
-          # not fine once a cross-table atomicity test came in.
-          set +e
-          for migration in $(ls migrations/*.sql 2>/dev/null | grep -v '\.down\.sql$' | sort); do
-            if psql -h "${PG_HOST}" -U postgres -d molecule -v ON_ERROR_STOP=1 \
-                  -f "$migration" >/dev/null 2>&1; then
-              echo "✓ $(basename "$migration")"
-            else
-              echo "⊘ $(basename "$migration") (skipped — see comment in workflow)"
-            fi
-          done
-          set -e
-
-          # Sanity: the delegations + workspaces + activity_logs tables
-          # MUST exist for the integration tests to be meaningful. Hard-
-          # fail if any didn't land — that would be a real regression we
-          # want loud.
-          for tbl in delegations workspaces activity_logs pending_uploads; do
-            if ! psql -h "${PG_HOST}" -U postgres -d molecule -tA \
-                -c "SELECT 1 FROM information_schema.tables WHERE table_name = '$tbl'" \
-                | grep -q 1; then
-              echo "::error::$tbl table missing after migration replay — handler integration tests would be meaningless"
-              exit 1
-            fi
-            echo "✓ $tbl table present"
-          done
-
-      - if: needs.detect-changes.outputs.handlers == 'true'
-        name: Run integration tests
-        run: |
-          # INTEGRATION_DB_URL is exported by the start-postgres step;
-          # points at the per-run bridge IP, not 127.0.0.1, so concurrent
-          # workflow runs don't fight over a host-net 5432 port.
-          go test -tags=integration -timeout 5m -v ./internal/handlers/ -run "^TestIntegration_"
-
-      - if: failure() && needs.detect-changes.outputs.handlers == 'true'
-        name: Diagnostic dump on failure
-        env:
-          PGPASSWORD: test
-        run: |
-          echo "::group::postgres container status"
-          docker ps -a --filter "name=${PG_NAME}" --format '{{.Status}} {{.Names}}' || true
-          docker logs "${PG_NAME}" 2>&1 | tail -50 || true
-          echo "::endgroup::"
-          echo "::group::delegations table state"
-          psql -h "${PG_HOST}" -U postgres -d molecule -c "SELECT * FROM delegations LIMIT 50;" || true
-          echo "::endgroup::"
-
-      - if: always() && needs.detect-changes.outputs.handlers == 'true'
-        name: Stop sibling Postgres
-        working-directory: .
-        run: |
-          # always() so containers don't leak when migrations or tests
-          # fail. The cleanup is best-effort: if the container is
-          # already gone (e.g. concurrent rerun race), don't fail the job.
-          docker rm -f "${PG_NAME}" >/dev/null 2>&1 || true
-          echo "Cleaned up ${PG_NAME}"
-
@@ -1,248 +0,0 @@
-name: Harness Replays
-
-# Boots tests/harness (production-shape compose topology with TenantGuard,
-# /cp/* proxy, canvas proxy, real production Dockerfile.tenant) and runs
-# every replay under tests/harness/replays/. Fails the PR if any replay
-# fails.
-#
-# Why this exists: 2026-04-30 we shipped #2398 which added /buildinfo as
-# a public route in router.go but forgot to add it to TenantGuard's
-# allowlist. The handler-level test in buildinfo_test.go constructed a
-# minimal gin engine without TenantGuard — green. The harness's
-# buildinfo-stale-image.sh replay would have caught it (cf-proxy doesn't
-# inject X-Molecule-Org-Id, so the curl path is identical to production's
-# redeploy verifier), but no one ran the harness pre-merge. The bug
-# shipped; the redeploy verifier silently soft-warned every tenant as
-# "unreachable" for ~1 day before being noticed.
-#
-# This gate makes "did you actually run the harness?" a CI invariant
-# instead of a memory-discipline thing.
-#
-# Trigger model — match e2e-api.yml: always FIRES on push/pull_request
-# to staging+main, real work is gated per-step on detect-changes output.
-# One job → one check run → branch-protection-clean (the SKIPPED-in-set
-# trap from PR #2264 is documented in e2e-api.yml's e2e-api job comment).
-
-on:
-  push:
-    branches: [main, staging]
-    paths:
-      - 'workspace-server/**'
-      - 'canvas/**'
-      - 'tests/harness/**'
-      - '.github/workflows/harness-replays.yml'
-  pull_request:
-    branches: [main, staging]
-    paths:
-      - 'workspace-server/**'
-      - 'canvas/**'
-      - 'tests/harness/**'
-      - '.github/workflows/harness-replays.yml'
-  workflow_dispatch:
-  merge_group:
-    types: [checks_requested]
-
-concurrency:
-  # Per-SHA grouping. Per-ref kept hitting the auto-promote-staging
-  # cancellation deadlock — see e2e-api.yml's concurrency block for
-  # the 2026-04-28 incident that codified this pattern.
-  group: harness-replays-${{ github.event.pull_request.head.sha || github.sha }}
-  cancel-in-progress: false
-
-jobs:
-  detect-changes:
-    runs-on: ubuntu-latest
-    outputs:
-      run: ${{ steps.decide.outputs.run }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - id: decide
-        run: |
-          # workflow_dispatch: always run (manual trigger)
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-            echo "run=true" >> "$GITHUB_OUTPUT"
-            echo "debug=manual-trigger" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          # Determine the base commit to diff against.
-          # For pull_request: use base.sha (the merge-base with main/staging).
-          # For push: use github.event.before (the previous tip of the branch).
-          # Fallback for new branches (all-zeros SHA): run everything.
-          if [ "${{ github.event_name }}" = "pull_request" ] && \
-             [ -n "${{ github.event.pull_request.base.sha }}" ]; then
-            BASE="${{ github.event.pull_request.base.sha }}"
-          elif [ -n "${{ github.event.before }}" ] && \
-               ! echo "${{ github.event.before }}" | grep -qE '^0+$'; then
-            BASE="${{ github.event.before }}"
-          else
-            # New branch or github.event.before unavailable — run everything.
-            echo "run=true" >> "$GITHUB_OUTPUT"
-            echo "debug=new-branch-fallback" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-
-          # GitHub Actions and Gitea Actions both expose github.sha for HEAD.
-          DIFF=$(git diff --name-only "$BASE" "${{ github.sha }}" 2>/dev/null)
-          echo "debug=diff-base=$BASE diff-files=$DIFF" >> "$GITHUB_OUTPUT"
-
-          if echo "$DIFF" | grep -qE '^workspace-server/|^canvas/|^tests/harness/|^.github/workflows/harness-replays\.yml$'; then
-            echo "run=true" >> "$GITHUB_OUTPUT"
-          else
-            echo "run=false" >> "$GITHUB_OUTPUT"
-          fi
-
-  # ONE job that always runs. Real work is gated per-step on
-  # detect-changes.outputs.run so an unrelated PR (e.g. doc-only
-  # change to molecule-controlplane wired here later) emits the
-  # required check without spending CI cycles. Single-job pattern
-  # matches e2e-api.yml — see that workflow's comment for why a
-  # job-level `if: false` would block branch protection via the
-  # SKIPPED-in-set bug.
-  harness-replays:
-    needs: detect-changes
-    name: Harness Replays
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    steps:
-      - name: No-op pass (paths filter excluded this commit)
-        if: needs.detect-changes.outputs.run != 'true'
-        run: |
-          echo "No workspace-server / canvas / tests/harness / workflow changes — Harness Replays gate satisfied without running."
-          echo "::notice::Harness Replays no-op pass (paths filter excluded this commit)."
-          echo "::notice::Debug: ${{ needs.detect-changes.outputs.debug }}"
-
-      - if: needs.detect-changes.outputs.run == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      # Log what files were detected so future failures include the diff.
-      - name: Log detected changes
-        if: needs.detect-changes.outputs.run == 'true'
-        run: |
-          echo "::notice::detect-changes debug: ${{ needs.detect-changes.outputs.debug }}"
-
-      # github-app-auth sibling-checkout removed 2026-05-07 (#157):
-      # the plugin was dropped + Dockerfile.tenant no longer COPYs it.
-
-      # Pre-clone manifest deps before docker compose builds the tenant
-      # image (Task #173 followup — same pattern as
-      # publish-workspace-server-image.yml's "Pre-clone manifest deps"
-      # step).
-      #
-      # Why pre-clone here too: tests/harness/compose.yml builds tenant-alpha
-      # and tenant-beta from workspace-server/Dockerfile.tenant with
-      # context=../.. (repo root). That Dockerfile expects
-      # .tenant-bundle-deps/{workspace-configs-templates,org-templates,plugins}
-      # to be present at build context root (post-#173 it COPYs from there
-      # instead of running an in-image clone — the in-image clone failed
-      # with "could not read Username for https://git.moleculesai.app"
-      # because there's no auth path inside the build sandbox).
-      #
-      # Without this step harness-replays fails before any replay runs,
-      # with `failed to calculate checksum of ref ...
-      # "/.tenant-bundle-deps/plugins": not found`. Caught by run #892
-      # (main, 2026-05-07T20:28:53Z) and run #964 (staging — same
-      # symptom, different root cause: staging still has the in-image
-      # clone path, hits the auth error directly).
-      #
-      # 2026-05-08 sub-finding (#192): the clone step ALSO fails when
-      # any referenced workspace-template repo is private and the
-      # AUTO_SYNC_TOKEN bearer (devops-engineer persona) lacks read
-      # access. Root cause: 5 of 9 workspace-template repos
-      # (openclaw, codex, crewai, deepagents, gemini-cli) had been
-      # marked private with no team grant. Resolution: flipped them
-      # to public per `feedback_oss_first_repo_visibility_default`
-      # (the OSS surface should be public). Layer-3 (customer-private +
-      # marketplace third-party repos) tracked separately in
-      # internal#102.
-      #
-      # Token shape matches publish-workspace-server-image.yml: AUTO_SYNC_TOKEN
-      # is the devops-engineer persona PAT, NOT the founder PAT (per
-      # `feedback_per_agent_gitea_identity_default`). clone-manifest.sh
-      # embeds it as basic-auth for the duration of the clones and strips
-      # .git directories — the token never enters the resulting image.
-      - name: Pre-clone manifest deps
-        if: needs.detect-changes.outputs.run == 'true'
-        env:
-          MOLECULE_GITEA_TOKEN: ${{ secrets.AUTO_SYNC_TOKEN }}
-        run: |
-          set -euo pipefail
-          if [ -z "${MOLECULE_GITEA_TOKEN}" ]; then
-            echo "::error::AUTO_SYNC_TOKEN secret is empty — register the devops-engineer persona PAT in repo Actions secrets"
-            exit 1
-          fi
-          mkdir -p .tenant-bundle-deps
-          bash scripts/clone-manifest.sh \
-            manifest.json \
-            .tenant-bundle-deps/workspace-configs-templates \
-            .tenant-bundle-deps/org-templates \
-            .tenant-bundle-deps/plugins
-          # Sanity-check counts so a silent partial clone fails fast
-          # instead of producing a half-empty image.
-          ws_count=$(find .tenant-bundle-deps/workspace-configs-templates -mindepth 1 -maxdepth 1 -type d | wc -l)
-          org_count=$(find .tenant-bundle-deps/org-templates -mindepth 1 -maxdepth 1 -type d | wc -l)
-          plugins_count=$(find .tenant-bundle-deps/plugins -mindepth 1 -maxdepth 1 -type d | wc -l)
-          echo "Cloned: ws=$ws_count org=$org_count plugins=$plugins_count"
-
-      - name: Install Python deps for replays
-        # peer-discovery-404 (and future replays) eval Python against the
-        # running tenant — importing workspace/a2a_client.py pulls in
-        # httpx. tests/harness/requirements.txt holds just the HTTP-client
-        # surface to keep CI install fast (~3s) vs the full
-        # workspace/requirements.txt (~30s).
-        if: needs.detect-changes.outputs.run == 'true'
-        run: pip install -r tests/harness/requirements.txt
-
-      - name: Run all replays against the harness
-        # run-all-replays.sh: boot via up.sh → seed via seed.sh → run
-        # every replays/*.sh → tear down via down.sh on EXIT (trap).
-        # Non-zero exit on any replay failure.
-        #
-        # KEEP_UP=1: without this, the script's trap-on-EXIT tears
-        # down containers immediately on failure, leaving the dump
-        # step below with nothing to dump (verified on PR #2410's
-        # first run — tenant became unhealthy, trap fired, dump
-        # step saw empty containers). Keeping them up lets the
-        # failure path collect tenant/cp-stub/cf-proxy logs. The
-        # always-run "Force teardown" step does the actual cleanup.
-        if: needs.detect-changes.outputs.run == 'true'
-        working-directory: tests/harness
-        env:
-          KEEP_UP: "1"
-        run: ./run-all-replays.sh
-
-      - name: Dump compose logs on failure
-        # SECRETS_ENCRYPTION_KEY: docker compose validates the entire compose
-        # file even for read-only `logs` calls. up.sh generates a per-run key
-        # and exports it to its OWN shell — this step runs in a fresh shell
-        # that wouldn't see it, so without a placeholder the validate step
-        # errors before logs print (verified against PR #2492's first run:
-        # "required variable SECRETS_ENCRYPTION_KEY is missing a value").
-        # A placeholder is fine — we're only reading log streams, not booting.
-        if: failure() && needs.detect-changes.outputs.run == 'true'
-        working-directory: tests/harness
-        env:
-          SECRETS_ENCRYPTION_KEY: dump-logs-placeholder
-        run: |
-          echo "=== docker compose ps ==="
-          docker compose -f compose.yml ps || true
-          echo "=== tenant-alpha logs ==="
-          docker compose -f compose.yml logs tenant-alpha || true
-          echo "=== tenant-beta logs ==="
-          docker compose -f compose.yml logs tenant-beta || true
-          echo "=== cp-stub logs ==="
-          docker compose -f compose.yml logs cp-stub || true
-          echo "=== cf-proxy logs ==="
-          docker compose -f compose.yml logs cf-proxy || true
-          echo "=== postgres-alpha logs (last 100) ==="
-          docker compose -f compose.yml logs --tail 100 postgres-alpha || true
-          echo "=== postgres-beta logs (last 100) ==="
-          docker compose -f compose.yml logs --tail 100 postgres-beta || true
-
-      - name: Force teardown
-        # We pass KEEP_UP=1 to run-all-replays.sh so the dump step
-        # above sees real containers — that means we own teardown
-        # explicitly here. Always run.
-        if: always() && needs.detect-changes.outputs.run == 'true'
-        working-directory: tests/harness
-        run: ./down.sh || true
@@ -1,94 +0,0 @@
-name: Lint curl status-code capture
-
-# Pins the workflow-bash anti-pattern that produced "HTTP 000000" on the
-# 2026-05-04 redeploy-tenants-on-main run for sha 2b862f6:
-#
-#   HTTP_CODE=$(curl ... -w '%{http_code}' ... || echo "000")
-#
-# When curl exits non-zero (connection reset → 56, --fail-with-body 4xx/5xx
-# → 22), the `-w '%{http_code}'` already wrote a status to stdout — usually
-# "000" for connection failures or the actual code for HTTP errors. The
-# `|| echo "000"` then fires AND appends ANOTHER "000" to the captured
-# stdout, producing values like "000000" or "409000" that fail string
-# comparisons against "200" while looking superficially right.
-#
-# Same class of bug the synth-E2E §7c gate hit twice (PRs #2779/#2783 +
-# #2797). Memory: feedback_curl_status_capture_pollution.md.
-#
-# Fix shape (route -w into a tempfile so curl's exit code can't pollute):
-#
-#   set +e
-#   curl ... -w '%{http_code}' >code.txt 2>/dev/null
-#   set -e
-#   HTTP_CODE=$(cat code.txt 2>/dev/null)
-#   [ -z "$HTTP_CODE" ] && HTTP_CODE="000"
-
-on:
-  pull_request:
-    paths: ['.github/workflows/**']
-  push:
-    branches: [main, staging]
-    paths: ['.github/workflows/**']
-  merge_group:
-    types: [checks_requested]
-
-jobs:
-  scan:
-    name: Scan workflows for curl status-capture pollution
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - name: Find curl ... -w '%{http_code}' ... || echo "000" subshells
-        run: |
-          set -uo pipefail
-          # Multi-line aware: look for `$(curl ... -w '%{http_code}' ... || echo "000")`
-          # subshell where the entire command-substitution wraps a curl that
-          # ends with `|| echo "000"`. Must distinguish from the SAFE shape
-          # `$(cat tempfile 2>/dev/null || echo "000")` — `cat` with a missing
-          # tempfile produces empty stdout, no pollution.
-          python3 <<'PY'
-          import os, re, sys, glob
-
-          BAD_FILES = []
-
-          # Match the buggy substitution across newlines: $(curl ... -w '%{http_code}' ... || echo "000")
-          # The `\\n` is the bash line-continuation that lets curl flags span lines.
-          # We collapse continuation lines first, then look for the single-line bad pattern.
-          PATTERN = re.compile(
-              r'\$\(\s*curl\b[^)]*-w\s*[\'"]%\{http_code\}[\'"][^)]*\|\|\s*echo\s+"000"\s*\)',
-              re.DOTALL,
-          )
-
-          # Self-skip: this lint workflow contains the literal anti-pattern in
-          # its own docstring — that's intentional, not a bug.
-          SELF = ".github/workflows/lint-curl-status-capture.yml"
-
-          for f in sorted(glob.glob(".github/workflows/*.yml")):
-              if f == SELF:
-                  continue
-              with open(f) as fh:
-                  content = fh.read()
-              # Collapse bash line-continuations (\\\n + leading whitespace)
-              # into a single logical line so the regex can see the full
-              # curl invocation as one chunk.
-              flat = re.sub(r'\\\s*\n\s*', ' ', content)
-              for m in PATTERN.finditer(flat):
-                  BAD_FILES.append((f, m.group(0)[:120]))
-
-          if not BAD_FILES:
-              print("✓ No curl-status-capture pollution patterns detected")
-              sys.exit(0)
-
-          print(f"::error::Found {len(BAD_FILES)} curl-status-capture pollution site(s):")
-          for f, snippet in BAD_FILES:
-              print(f"::error file={f}::Curl status-capture pollution: '|| echo \"000\"' inside a $(curl ... -w '%{{http_code}}' ...) subshell. On non-2xx or connection failure, curl's -w writes a status, then exits non-zero, then the || echo appends another '000' — producing 'HTTP 000000' or '409000' that fails comparisons silently. Fix: route -w into a tempfile so the exit code can't pollute stdout. See memory feedback_curl_status_capture_pollution.md.")
-              print(f"   matched: {snippet}…")
-          print()
-          print("Fix template:")
-          print('  set +e')
-          print('  curl ... -w \'%{http_code}\' >code.txt 2>/dev/null')
-          print('  set -e')
-          print('  HTTP_CODE=$(cat code.txt 2>/dev/null)')
-          print('  [ -z "$HTTP_CODE" ] && HTTP_CODE="000"')
-          sys.exit(1)
-          PY
@@ -1,121 +0,0 @@
-name: publish-canvas-image
-
-# Builds and pushes the canvas Docker image to GHCR whenever a commit lands
-# on main that touches canvas code. Previously canvas changes were visible in
-# CI (npm run build passed) but the live container was never updated —
-# operators had to manually run `docker compose build canvas` each time.
-#
-# Mirror of publish-platform-image.yml, adapted for the Next.js canvas layer.
-# See that workflow for inline notes on macOS Keychain isolation and QEMU.
-
-on:
-  push:
-    branches: [main]
-    paths:
-      # Only rebuild when canvas source changes — saves GHA minutes on
-      # platform-only / docs-only / MCP-only merges.
-      - 'canvas/**'
-      - '.github/workflows/publish-canvas-image.yml'
-  # Manual trigger: use after a non-canvas merge that still needs a fresh
-  # image (e.g. a Dockerfile change lives outside the canvas/ tree).
-  workflow_dispatch:
-    inputs:
-      platform_url:
-        description: 'NEXT_PUBLIC_PLATFORM_URL baked into the bundle (default: http://localhost:8080)'
-        required: false
-        default: ''
-      ws_url:
-        description: 'NEXT_PUBLIC_WS_URL baked into the bundle (default: ws://localhost:8080/ws)'
-        required: false
-        default: ''
-
-permissions:
-  contents: read
-  packages: write  # required to push to ghcr.io/${{ github.repository_owner }}/*
-
-env:
-  IMAGE_NAME: ghcr.io/molecule-ai/canvas
-
-jobs:
-  build-and-push:
-    name: Build & push canvas image
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Log in to GHCR
-        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
-        with:
-          registry: ghcr.io
-          username: ${{ github.actor }}
-          password: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd # v4.0.0
-
-      # Health check: verify Docker daemon is accessible before attempting any
-      # build steps. This fails loudly at step 1 when the runner's docker.sock
-      # is inaccessible rather than silently continuing to the build step
-      # where docker build fails deep in ECR auth with a cryptic error.
-      - name: Verify Docker daemon access
-        run: |
-          set -euo pipefail
-          echo "::group::Docker daemon health check"
-          docker info 2>&1 | head -5 || {
-            echo "::error::Docker daemon is not accessible at /var/run/docker.sock"
-            echo "::error::Check: (1) daemon running, (2) runner user in docker group, (3) sock perms 660+"
-            exit 1
-          }
-          echo "Docker daemon OK"
-          echo "::endgroup::"
-
-      - name: Compute tags
-        id: tags
-        shell: bash
-        run: |
-          echo "sha=${GITHUB_SHA::7}" >> "$GITHUB_OUTPUT"
-
-      - name: Resolve build args
-        id: build_args
-        # Priority: workflow_dispatch input > repo secret > hardcoded default.
-        # NEXT_PUBLIC_* env vars are baked into the JS bundle at build time by
-        # Next.js — they cannot be changed at runtime without a full rebuild.
-        # For local docker-compose deployments the defaults (localhost:8080)
-        # work as-is; production deployments should set CANVAS_PLATFORM_URL
-        # and CANVAS_WS_URL as repository secrets.
-        #
-        # Inputs are passed via env vars (not direct ${{ }} interpolation) to
-        # prevent shell injection from workflow_dispatch string inputs.
-        shell: bash
-        env:
-          INPUT_PLATFORM_URL: ${{ github.event.inputs.platform_url }}
-          SECRET_PLATFORM_URL: ${{ secrets.CANVAS_PLATFORM_URL }}
-          INPUT_WS_URL: ${{ github.event.inputs.ws_url }}
-          SECRET_WS_URL: ${{ secrets.CANVAS_WS_URL }}
-        run: |
-          PLATFORM_URL="${INPUT_PLATFORM_URL:-${SECRET_PLATFORM_URL:-http://localhost:8080}}"
-          WS_URL="${INPUT_WS_URL:-${SECRET_WS_URL:-ws://localhost:8080/ws}}"
-
-          echo "platform_url=${PLATFORM_URL}" >> "$GITHUB_OUTPUT"
-          echo "ws_url=${WS_URL}" >> "$GITHUB_OUTPUT"
-
-      - name: Build & push canvas image to GHCR
-        uses: docker/build-push-action@bcafcacb16a39f128d818304e6c9c0c18556b85f # v7.1.0
-        with:
-          context: ./canvas
-          file: ./canvas/Dockerfile
-          platforms: linux/amd64
-          push: true
-          build-args: |
-            NEXT_PUBLIC_PLATFORM_URL=${{ steps.build_args.outputs.platform_url }}
-            NEXT_PUBLIC_WS_URL=${{ steps.build_args.outputs.ws_url }}
-          tags: |
-            ${{ env.IMAGE_NAME }}:latest
-            ${{ env.IMAGE_NAME }}:sha-${{ steps.tags.outputs.sha }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
-          labels: |
-            org.opencontainers.image.source=https://github.com/${{ github.repository }}
-            org.opencontainers.image.revision=${{ github.sha }}
-            org.opencontainers.image.description=Molecule AI canvas (Next.js 15 + React Flow)
@@ -1,207 +0,0 @@
-name: Railway pin audit (drift detection)
-
-# Daily audit of Railway env vars for drift-prone image-tag pins —
-# automation-cadence layer over the detection script + regression test
-# shipped in PR #2168 (#2001 closure).
-#
-# Background: on 2026-04-24 a stale `:staging-a14cf86` SHA pin in CP's
-# TENANT_IMAGE caused 3+ hours of E2E failure with the appearance that
-# "every fix didn't propagate" — really the tenant image was so old it
-# didn't read the env vars those fixes produced. The audit script
-# (scripts/ops/audit-railway-sha-pins.sh) flags drift; this workflow
-# runs the same check unattended on a daily cron.
-#
-# Cadence: once a day, 13:00 UTC (06:00 PT). Daily is the right
-# cadence for variables-tier config — Railway env var changes are
-# deliberate operator actions, low-frequency. Hourly would risk
-# Railway API rate-limit surprises and is overkill for the change rate.
-#
-# Issue-on-failure: drift triggers a priority-high issue, mirroring
-# .github/workflows/e2e-staging-sanity.yml's pattern. Drift is
-# medium-priority "config slipped, fix at next ops window," not
-# active-outage paging.
-#
-# Secret hardening: per feedback_schedule_vs_dispatch_secrets_hardening,
-# the schedule trigger HARD-FAILS on missing RAILWAY_AUDIT_TOKEN
-# (silent-success on schedule was the failure-mode class that bit the
-# team before; cron firing without checking anything is worse than no
-# cron). The workflow_dispatch trigger SOFT-SKIPS on missing secret so
-# an operator can dry-run the workflow shape during initial provisioning
-# without tripping a fake red.
-
-on:
-  schedule:
-    - cron: '0 13 * * *'
-  workflow_dispatch:
-
-concurrency:
-  group: railway-pin-audit
-  cancel-in-progress: false
-
-permissions:
-  issues: write
-  contents: read
-
-jobs:
-  audit:
-    name: Audit Railway env vars for drift-prone pins
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify RAILWAY_AUDIT_TOKEN present
-        # Schedule trigger: hard-fail when the secret is missing —
-        # otherwise the cron silently runs against the wrong scope (or
-        # exits 2 from the script and we issue-spam) without anyone
-        # noticing the token rot.
-        # Dispatch trigger: soft-skip — operator may be dry-running the
-        # workflow shape before provisioning the secret. Logged as a
-        # workflow notice, not a failure.
-        env:
-          RAILWAY_AUDIT_TOKEN: ${{ secrets.RAILWAY_AUDIT_TOKEN }}
-          EVENT_NAME: ${{ github.event_name }}
-        id: secret_check
-        run: |
-          set -euo pipefail
-          if [ -n "${RAILWAY_AUDIT_TOKEN:-}" ]; then
-            echo "have_secret=true" >> "$GITHUB_OUTPUT"
-            exit 0
-          fi
-          echo "have_secret=false" >> "$GITHUB_OUTPUT"
-          if [ "$EVENT_NAME" = "workflow_dispatch" ]; then
-            echo "::notice::RAILWAY_AUDIT_TOKEN not configured — soft-skipping (manual dispatch)"
-            exit 0
-          fi
-          echo "::error::RAILWAY_AUDIT_TOKEN secret missing — schedule trigger requires it. Provision the token (read-only \`variables\` scope on the molecule-platform Railway project) and store as repo secret RAILWAY_AUDIT_TOKEN."
-          exit 1
-
-      - name: Install Railway CLI
-        if: steps.secret_check.outputs.have_secret == 'true'
-        # Pinned hash matching the public install instructions; bump in
-        # tandem with the audit-script's documented Railway CLI version.
-        run: |
-          set -euo pipefail
-          curl -fsSL https://railway.com/install.sh | sh
-          # The installer drops the binary in ~/.railway/bin
-          echo "$HOME/.railway/bin" >> "$GITHUB_PATH"
-
-      - name: Verify Railway CLI authenticated
-        if: steps.secret_check.outputs.have_secret == 'true'
-        env:
-          RAILWAY_TOKEN: ${{ secrets.RAILWAY_AUDIT_TOKEN }}
-        run: |
-          set -euo pipefail
-          # `railway whoami` exits non-zero when the token is
-          # unauthenticated or doesn't have any project access.
-          if ! railway whoami >/dev/null 2>&1; then
-            echo "::error::Railway CLI failed to authenticate with RAILWAY_AUDIT_TOKEN — token may be revoked or scoped incorrectly"
-            exit 2
-          fi
-
-      - name: Link molecule-platform project
-        if: steps.secret_check.outputs.have_secret == 'true'
-        env:
-          RAILWAY_TOKEN: ${{ secrets.RAILWAY_AUDIT_TOKEN }}
-        # Project ID from reference_production_stack: molecule-platform
-        # / 7ccc8c68-61f4-42ab-9be5-586eeee11768. Linking is per-process,
-        # so we re-link in this CI shell (the audit script comment says
-        # it deliberately doesn't chdir for you because the linked
-        # project's identity matters).
-        run: |
-          set -euo pipefail
-          railway link --project 7ccc8c68-61f4-42ab-9be5-586eeee11768
-
-      - name: Run drift audit
-        if: steps.secret_check.outputs.have_secret == 'true'
-        id: audit
-        env:
-          RAILWAY_TOKEN: ${{ secrets.RAILWAY_AUDIT_TOKEN }}
-        run: |
-          set +e
-          bash scripts/ops/audit-railway-sha-pins.sh 2>&1 | tee /tmp/audit.log
-          rc=${PIPESTATUS[0]}
-          echo "rc=$rc" >> "$GITHUB_OUTPUT"
-          # Capture the audit log for the issue body.
-          {
-            echo 'log<<AUDIT_EOF'
-            cat /tmp/audit.log
-            echo 'AUDIT_EOF'
-          } >> "$GITHUB_OUTPUT"
-          # Exit codes from the script:
-          #   0 — no drift; workflow goes green
-          #   1 — drift detected; we'll file an issue and fail the run
-          #   2 — railway CLI unauthenticated / project unlinked; fail
-          # Anything else: also fail.
-          case "$rc" in
-            0) exit 0 ;;
-            1) echo "::warning::Drift-prone pin(s) detected — issue will be filed"; exit 1 ;;
-            2) echo "::error::Railway CLI auth/link failed mid-script — token or project ID drift"; exit 2 ;;
-            *) echo "::error::Unexpected audit rc=$rc"; exit 1 ;;
-          esac
-
-      - name: Open / update drift issue
-        if: failure() && steps.audit.outputs.rc == '1'
-        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
-        env:
-          AUDIT_LOG: ${{ steps.audit.outputs.log }}
-        with:
-          script: |
-            const title = "🚨 Railway env-var drift detected";
-            const runURL = `https://github.com/${context.repo.owner}/${context.repo.repo}/actions/runs/${context.runId}`;
-            const body =
-              `Daily Railway pin audit found drift-prone image-tag pins in the molecule-platform Railway project.\n\n` +
-              `**What this means:** an env var (likely on \`controlplane\`) is pinned to a SHA-shaped or semver tag instead of a floating tag. ` +
-              `Same pattern that caused the 2026-04-24 TENANT_IMAGE incident — fix-PRs land but the running service doesn't pick them up.\n\n` +
-              `**Recovery:** open the Railway dashboard, replace the flagged value with a floating tag (\`:staging-latest\`, \`:main\`) unless the pin is intentional and documented in the ops runbook.\n\n` +
-              `**Audit output:**\n\n\`\`\`\n${process.env.AUDIT_LOG || '(log unavailable)'}\n\`\`\`\n\n` +
-              `Run: ${runURL}\n\n` +
-              `Closes automatically when a subsequent daily run reports clean.`;
-
-            const { data: existing } = await github.rest.issues.listForRepo({
-              owner: context.repo.owner, repo: context.repo.repo,
-              state: 'open', labels: 'railway-drift',
-            });
-            const match = existing.find(i => i.title === title);
-            if (match) {
-              await github.rest.issues.createComment({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: match.number,
-                body: `Still drifting. ${runURL}\n\n\`\`\`\n${process.env.AUDIT_LOG || '(log unavailable)'}\n\`\`\``,
-              });
-            } else {
-              await github.rest.issues.create({
-                owner: context.repo.owner, repo: context.repo.repo,
-                title, body,
-                labels: ['railway-drift', 'bug', 'priority-high'],
-              });
-            }
-
-      - name: Close stale drift issue on clean run
-        # When a previously-flagged drift gets fixed by an operator,
-        # the next daily run goes green. Close any open `railway-drift`
-        # issue with a confirmation comment so the queue doesn't carry
-        # stale ones.
-        if: success() && steps.audit.outputs.rc == '0'
-        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
-        with:
-          script: |
-            const runURL = `https://github.com/${context.repo.owner}/${context.repo.repo}/actions/runs/${context.runId}`;
-            const { data: existing } = await github.rest.issues.listForRepo({
-              owner: context.repo.owner, repo: context.repo.repo,
-              state: 'open', labels: 'railway-drift',
-            });
-            for (const issue of existing) {
-              await github.rest.issues.createComment({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: issue.number,
-                body: `Daily audit clean — drift resolved. ${runURL}`,
-              });
-              await github.rest.issues.update({
-                owner: context.repo.owner, repo: context.repo.repo,
-                issue_number: issue.number,
-                state: 'closed',
-                state_reason: 'completed',
-              });
-            }
@@ -1,400 +0,0 @@
-name: redeploy-tenants-on-main
-
-# Auto-refresh prod tenant EC2s after every main merge.
-#
-# Why this workflow exists: publish-workspace-server-image builds and
-# pushes a new platform-tenant :<sha> to ECR on every merge to main,
-# but running tenants pulled their image once at boot and never re-pull.
-# Users see stale code indefinitely.
-#
-# This workflow closes the gap by calling the control-plane admin
-# endpoint that performs a canary-first, batched, health-gated rolling
-# redeploy across every live tenant. Implemented in molecule-ai/
-# molecule-controlplane as POST /cp/admin/tenants/redeploy-fleet
-# (feat/tenant-auto-redeploy, landing alongside this workflow).
-#
-# Registry: ECR (153263036946.dkr.ecr.us-east-2.amazonaws.com/
-# molecule-ai/platform-tenant). GHCR was retired 2026-05-07 during the
-# Gitea suspension migration. The canary-verify.yml promote step now
-# uses the same redeploy-fleet endpoint (fixes the silent-GHCR gap).
-#
-# Runtime ordering:
-#   1. publish-workspace-server-image completes → new :staging-<sha> in ECR.
-#   2. This workflow fires via workflow_run, calls redeploy-fleet with
-#      target_tag=staging-<sha>. No CDN propagation wait needed —
-#      ECR image manifest is consistent immediately after push.
-#   3. Calls redeploy-fleet with canary_slug (if set) and a soak
-#      period. Canary proves the image boots; batches follow.
-#   4. Any failure aborts the rollout and leaves older tenants on the
-#      prior image — safer default than half-and-half state.
-#
-# Rollback path: re-run this workflow with a specific SHA pinned via
-# the workflow_dispatch input. That calls redeploy-fleet with
-# target_tag=<sha>, re-pulling the older image on every tenant.
-
-on:
-  workflow_run:
-    workflows: ['publish-workspace-server-image']
-    types: [completed]
-    branches: [main]
-  workflow_dispatch:
-    inputs:
-      target_tag:
-        # Empty default → auto-trigger and dispatch-without-input both
-        # resolve to `staging-<short_head_sha>` (the digest publish-image
-        # just pushed). Pre-fix this defaulted to 'latest', which only
-        # gets retagged by canary-verify's promote-to-latest job — and
-        # that job soft-skips when CANARY_TENANT_URLS is unset (the
-        # current state, until Phase 2 canary fleet is live). Result:
-        # `:latest` had been pinned to a 4-day-old digest (2026-04-28)
-        # while every main push pushed fresh `staging-<sha>` images;
-        # every prod redeploy pulled the stale `:latest` and the verify
-        # step correctly flagged 3/3 tenants STALE. Pulling the
-        # just-published `staging-<sha>` directly skips the dead retag
-        # path. When canary fleet is real, this workflow should chain
-        # on canary-verify completion (workflow_run from canary-verify),
-        # not publish-image — separate, smaller PR.
-        description: 'Tenant image tag to deploy (e.g. "latest", "staging-a59f1a6c"). Empty = auto staging-<head_sha>.'
-        required: false
-        type: string
-        default: ''
-      canary_slug:
-        description: 'Tenant slug to deploy first + soak (empty = skip canary, fan out immediately).'
-        required: false
-        type: string
-        # Must be an actual prod tenant slug (current: hongming,
-        # chloe-dong, reno-stars). The previous default 'hongmingwang'
-        # didn't match any tenant — CP soft-skipped the missing canary
-        # and the fleet rolled out without the soak gate, defeating the
-        # whole point of canary-first.
-        default: 'hongming'
-      soak_seconds:
-        description: 'Seconds to wait after canary before fanning out.'
-        required: false
-        type: string
-        default: '60'
-      batch_size:
-        description: 'How many tenants SSM redeploys in parallel per batch.'
-        required: false
-        type: string
-        default: '3'
-      dry_run:
-        description: 'Plan only — do not actually redeploy.'
-        required: false
-        type: boolean
-        default: false
-
-permissions:
-  contents: read
-  # No write scopes needed — the workflow hits an external CP endpoint,
-  # not the GitHub API.
-
-# Serialize redeploys so two rapid main pushes' redeploys don't overlap
-# and cause confusing per-tenant SSM state. Without this, GitHub's
-# implicit workflow_run queueing would *probably* serialize them, but
-# the explicit block makes the invariant defensible. Mirrors the
-# concurrency block on redeploy-tenants-on-staging.yml for shape parity.
-#
-# cancel-in-progress: false → aborting a half-rolled-out fleet would
-# leave tenants stuck on whatever image they happened to be on when
-# cancelled. Better to finish the in-flight rollout before starting
-# the next one.
-concurrency:
-  group: redeploy-tenants-on-main
-  cancel-in-progress: false
-
-jobs:
-  redeploy:
-    # Skip the auto-trigger if publish-workspace-server-image didn't
-    # actually succeed. workflow_run fires on any completion state; we
-    # don't want to redeploy against a half-built image.
-    if: |
-      github.event_name == 'workflow_dispatch' ||
-      (github.event_name == 'workflow_run' && github.event.workflow_run.conclusion == 'success')
-    runs-on: ubuntu-latest
-    timeout-minutes: 25
-    steps:
-      - name: Note on ECR propagation
-        # ECR image manifests are consistent immediately after push — no
-        # CDN cache to wait for. The old GHCR-based workflow had a 30s
-        # sleep to avoid race conditions; ECR makes that unnecessary.
-        run: echo "ECR image available immediately after push — proceeding."
-
-      - name: Compute target tag
-        id: tag
-        # Resolution order:
-        #   1. Operator-supplied input (workflow_dispatch with explicit
-        #      tag) → used verbatim. Lets ops pin `latest` for emergency
-        #      rollback to last canary-verified digest, or pin a specific
-        #      `staging-<sha>` to roll back to a known-good build.
-        #   2. Default → `staging-<short_head_sha>`. The just-published
-        #      digest. Bypasses the `:latest` retag path that's currently
-        #      dead (canary-verify soft-skips without canary fleet, so
-        #      the only thing retagging `:latest` today is the manual
-        #      promote-latest.yml — last run 2026-04-28). Auto-trigger
-        #      from workflow_run uses workflow_run.head_sha; manual
-        #      dispatch with no input falls through to github.sha.
-        env:
-          INPUT_TAG: ${{ inputs.target_tag }}
-          HEAD_SHA: ${{ github.event.workflow_run.head_sha || github.sha }}
-        run: |
-          set -euo pipefail
-          if [ -n "${INPUT_TAG:-}" ]; then
-            echo "target_tag=$INPUT_TAG" >> "$GITHUB_OUTPUT"
-            echo "Using operator-pinned tag: $INPUT_TAG"
-          else
-            SHORT="${HEAD_SHA:0:7}"
-            echo "target_tag=staging-$SHORT" >> "$GITHUB_OUTPUT"
-            echo "Using auto tag: staging-$SHORT (head_sha=$HEAD_SHA)"
-          fi
-
-      - name: Call CP redeploy-fleet
-        # CP_ADMIN_API_TOKEN must be set as a repo/org secret on
-        # molecule-ai/molecule-core, matching the staging/prod CP's
-        # CP_ADMIN_API_TOKEN env. Stored in Railway, mirrored to this
-        # repo's secrets for CI.
-        env:
-          CP_URL: ${{ vars.CP_URL || 'https://api.moleculesai.app' }}
-          CP_ADMIN_API_TOKEN: ${{ secrets.CP_ADMIN_API_TOKEN }}
-          TARGET_TAG: ${{ steps.tag.outputs.target_tag }}
-          CANARY_SLUG: ${{ inputs.canary_slug || 'hongming' }}
-          SOAK_SECONDS: ${{ inputs.soak_seconds || '60' }}
-          BATCH_SIZE: ${{ inputs.batch_size || '3' }}
-          DRY_RUN: ${{ inputs.dry_run || false }}
-        run: |
-          set -euo pipefail
-
-          if [ -z "${CP_ADMIN_API_TOKEN:-}" ]; then
-            echo "::error::CP_ADMIN_API_TOKEN secret not set — skipping redeploy"
-            echo "::notice::Set CP_ADMIN_API_TOKEN in repo secrets to enable auto-redeploy."
-            exit 1
-          fi
-
-          BODY=$(jq -nc \
-            --arg tag "$TARGET_TAG" \
-            --arg canary "$CANARY_SLUG" \
-            --argjson soak "$SOAK_SECONDS" \
-            --argjson batch "$BATCH_SIZE" \
-            --argjson dry "$DRY_RUN" \
-            '{
-              target_tag: $tag,
-              canary_slug: $canary,
-              soak_seconds: $soak,
-              batch_size: $batch,
-              dry_run: $dry
-            }')
-
-          echo "POST $CP_URL/cp/admin/tenants/redeploy-fleet"
-          echo "  body: $BODY"
-
-          HTTP_RESPONSE=$(mktemp)
-          HTTP_CODE_FILE=$(mktemp)
-          # Route -w into its own tempfile so curl's exit code (e.g. 56
-          # on connection-reset, 22 on --fail-with-body 4xx/5xx) can't
-          # pollute the captured stdout. The previous inline-substitution
-          # shape produced "000000" on connection reset (curl wrote
-          # "000" via -w, then the inline echo-fallback appended another
-          # "000") — caught on the 2026-05-04 redeploy of sha 2b862f6.
-          # set +e/-e keeps the non-zero curl exit from tripping the
-          # outer pipeline. See lint-curl-status-capture.yml for the
-          # CI gate that pins this fix shape.
-          set +e
-          curl -sS -o "$HTTP_RESPONSE" -w '%{http_code}' \
-            -m 1200 \
-            -H "Authorization: Bearer $CP_ADMIN_API_TOKEN" \
-            -H "Content-Type: application/json" \
-            -X POST "$CP_URL/cp/admin/tenants/redeploy-fleet" \
-            -d "$BODY" >"$HTTP_CODE_FILE"
-          set -e
-          # Stderr from curl (e.g. dial errors with -sS) goes to the runner
-          # log so operators can see WHY a connection failed. Stdout is
-          # captured to $HTTP_CODE_FILE because that's where -w writes.
-          HTTP_CODE=$(cat "$HTTP_CODE_FILE" 2>/dev/null || echo "000")
-          [ -z "$HTTP_CODE" ] && HTTP_CODE="000"
-
-          echo "HTTP $HTTP_CODE"
-          cat "$HTTP_RESPONSE" | jq . || cat "$HTTP_RESPONSE"
-
-          # Pretty-print per-tenant results in the job summary so
-          # ops can see which tenants were redeployed without drilling
-          # into the raw response.
-          {
-            echo "## Tenant redeploy fleet"
-            echo ""
-            echo "**Target tag:** \`$TARGET_TAG\`"
-            echo "**Canary:** \`$CANARY_SLUG\` (soak ${SOAK_SECONDS}s)"
-            echo "**Batch size:** $BATCH_SIZE"
-            echo "**Dry run:** $DRY_RUN"
-            echo "**HTTP:** $HTTP_CODE"
-            echo ""
-            echo "### Per-tenant result"
-            echo ""
-            echo '| Slug | Phase | SSM Status | Exit | Healthz | Error |'
-            echo '|------|-------|------------|------|---------|-------|'
-            jq -r '.results[]? | "| \(.slug) | \(.phase) | \(.ssm_status // "-") | \(.ssm_exit_code) | \(.healthz_ok) | \(.error // "-") |"' "$HTTP_RESPONSE" || true
-          } >> "$GITHUB_STEP_SUMMARY"
-
-          if [ "$HTTP_CODE" != "200" ]; then
-            echo "::error::redeploy-fleet returned HTTP $HTTP_CODE"
-            exit 1
-          fi
-          OK=$(jq -r '.ok' "$HTTP_RESPONSE")
-          if [ "$OK" != "true" ]; then
-            echo "::error::redeploy-fleet reported ok=false (see summary for which tenant halted the rollout)"
-            exit 1
-          fi
-          echo "::notice::Tenant fleet redeploy reported ssm_status=Success — verifying actual image roll on each tenant..."
-
-          # Stash the response for the verify step. $RUNNER_TEMP outlasts
-          # the step boundary; $HTTP_RESPONSE doesn't.
-          cp "$HTTP_RESPONSE" "$RUNNER_TEMP/redeploy-response.json"
-
-      - name: Verify each tenant /buildinfo matches published SHA
-        # ROOT FIX FOR #2395.
-        #
-        # `redeploy-fleet`'s `ssm_status=Success` means "the SSM RPC
-        # didn't error" — NOT "the new image is running on the tenant."
-        # `:latest` lives in the local Docker daemon's image cache; if
-        # the SSM document does `docker compose up -d` without an
-        # explicit `docker pull`, the daemon serves the previously-
-        # cached digest and the container restarts on stale code.
-        # 2026-04-30 incident: hongmingwang's tenant reported
-        # ssm_status=Success at 17:00:53Z but kept serving pre-501a42d7
-        # chat_files for 30+ min — the lazy-heal fix never reached the
-        # user despite green deploy + green redeploy.
-        #
-        # This step closes the gap by curling each tenant's /buildinfo
-        # endpoint (added in workspace-server/internal/buildinfo +
-        # /Dockerfile* GIT_SHA build-arg, this PR) and comparing the
-        # returned git_sha to the SHA the workflow expects. Mismatches
-        # fail the workflow, which is what `ok=true` should have
-        # guaranteed all along.
-        #
-        # When the redeploy was triggered by workflow_dispatch with a
-        # specific tag (target_tag != "latest"), the expected SHA may
-        # not equal ${{ github.sha }} — in that case we resolve via
-        # GHCR's manifest. For workflow_run (default :latest) the
-        # workflow_run.head_sha is the SHA that just published.
-        env:
-          EXPECTED_SHA: ${{ github.event.workflow_run.head_sha || github.sha }}
-          TARGET_TAG: ${{ steps.tag.outputs.target_tag }}
-          # Tenant subdomain template — slugs from the response are
-          # appended. Production CP issues `<slug>.moleculesai.app`;
-          # staging CP issues `<slug>.staging.moleculesai.app`. This
-          # workflow runs on main → prod CP → no `staging.` infix.
-          TENANT_DOMAIN: 'moleculesai.app'
-        run: |
-          set -euo pipefail
-
-          EXPECTED_SHORT="${EXPECTED_SHA:0:7}"
-          if [ "$TARGET_TAG" != "latest" ] \
-             && [ "$TARGET_TAG" != "$EXPECTED_SHA" ] \
-             && [ "$TARGET_TAG" != "staging-$EXPECTED_SHORT" ]; then
-            # workflow_dispatch with a pinned tag that isn't the head
-            # SHA — operator is rolling back / pinning. Skip the
-            # verification because we don't have the expected SHA in
-            # this context (would need to crane-inspect the GHCR
-            # manifest, which is a follow-up). Failing-open here is
-            # safe: the operator chose the tag deliberately.
-            #
-            # `staging-<short_head_sha>` IS verified — it's the new
-            # auto-trigger default (see Compute target tag step) and
-            # the digest under that tag SHOULD match EXPECTED_SHA.
-            echo "::notice::target_tag=$TARGET_TAG (operator-pinned) — skipping per-tenant SHA verification."
-            exit 0
-          fi
-
-          RESP="$RUNNER_TEMP/redeploy-response.json"
-          if [ ! -s "$RESP" ]; then
-            echo "::error::redeploy-response.json missing or empty — verify step ran without a response to read"
-            exit 1
-          fi
-
-          # Pull only successfully-redeployed tenants. Any tenant that
-          # halted the rollout already failed the previous step, so we
-          # don't double-count them here.
-          mapfile -t SLUGS < <(jq -r '.results[]? | select(.healthz_ok == true) | .slug' "$RESP")
-          if [ ${#SLUGS[@]} -eq 0 ]; then
-            echo "::warning::No tenants reported healthz_ok — nothing to verify"
-            exit 0
-          fi
-
-          echo "Verifying ${#SLUGS[@]} tenant(s) against EXPECTED_SHA=${EXPECTED_SHA:0:7}..."
-
-          # Two distinct failure modes — STALE (the #2395 bug class, hard-fail)
-          # vs UNREACHABLE (teardown race, soft-warn). See the staging variant's
-          # comment for the full rationale; same logic applies on prod even
-          # though prod has fewer ephemeral tenants — the asymmetry would be a
-          # gratuitous fork.
-          STALE_COUNT=0
-          UNREACHABLE_COUNT=0
-          STALE_LINES=()
-          UNREACHABLE_LINES=()
-          for slug in "${SLUGS[@]}"; do
-            URL="https://${slug}.${TENANT_DOMAIN}/buildinfo"
-            # 30s total: tenant just SSM-restarted, may still be coming
-            # up. Retry-on-empty rather than retry-on-status — we want
-            # to fail fast on "responded with wrong SHA", not "still
-            # warming up".
-            BODY=$(curl -sS --max-time 30 --retry 3 --retry-delay 5 --retry-connrefused "$URL" || true)
-            ACTUAL_SHA=$(echo "$BODY" | jq -r '.git_sha // ""' 2>/dev/null || echo "")
-            if [ -z "$ACTUAL_SHA" ]; then
-              UNREACHABLE_COUNT=$((UNREACHABLE_COUNT + 1))
-              UNREACHABLE_LINES+=("| $slug | (no /buildinfo response) | ${EXPECTED_SHA:0:7} | ⚠ unreachable (likely teardown race) |")
-              continue
-            fi
-            if [ "$ACTUAL_SHA" = "$EXPECTED_SHA" ]; then
-              echo "  $slug: ${ACTUAL_SHA:0:7} ✓"
-            else
-              STALE_COUNT=$((STALE_COUNT + 1))
-              STALE_LINES+=("| $slug | ${ACTUAL_SHA:0:7} | ${EXPECTED_SHA:0:7} | ❌ stale |")
-            fi
-          done
-
-          {
-            echo ""
-            echo "### Per-tenant /buildinfo verification"
-            echo ""
-            echo "Expected SHA: \`${EXPECTED_SHA:0:7}\`"
-            echo ""
-            if [ $STALE_COUNT -gt 0 ]; then
-              echo "**${STALE_COUNT} STALE tenant(s) — these did NOT pick up the new image despite ssm_status=Success:**"
-              echo ""
-              echo "| Slug | Actual /buildinfo SHA | Expected | Status |"
-              echo "|------|----------------------|----------|--------|"
-              for line in "${STALE_LINES[@]}"; do echo "$line"; done
-              echo ""
-            fi
-            if [ $UNREACHABLE_COUNT -gt 0 ]; then
-              echo "**${UNREACHABLE_COUNT} unreachable tenant(s) — likely teardown race (soft-warn, not failing):**"
-              echo ""
-              echo "| Slug | Actual /buildinfo SHA | Expected | Status |"
-              echo "|------|----------------------|----------|--------|"
-              for line in "${UNREACHABLE_LINES[@]}"; do echo "$line"; done
-              echo ""
-            fi
-            if [ $STALE_COUNT -eq 0 ] && [ $UNREACHABLE_COUNT -eq 0 ]; then
-              echo "All ${#SLUGS[@]} tenants returned matching SHA. ✓"
-            fi
-          } >> "$GITHUB_STEP_SUMMARY"
-
-          if [ $UNREACHABLE_COUNT -gt 0 ]; then
-            echo "::warning::$UNREACHABLE_COUNT tenant(s) unreachable post-redeploy. Likely benign teardown race — CP healthz monitor catches real outages."
-          fi
-
-          # Belt-and-suspenders sanity floor: same logic as the staging
-          # variant — see that file's comment for the full rationale.
-          # Floor only applies when fleet >= 4; below that, canary-verify
-          # is the actual gate.
-          TOTAL_VERIFIED=${#SLUGS[@]}
-          if [ $TOTAL_VERIFIED -ge 4 ] && [ $UNREACHABLE_COUNT -gt $((TOTAL_VERIFIED / 2)) ]; then
-            echo "::error::$UNREACHABLE_COUNT of $TOTAL_VERIFIED tenant(s) unreachable — exceeds 50% threshold on a fleet large enough that this signals a real outage, not teardown race."
-            exit 1
-          fi
-
-          if [ $STALE_COUNT -gt 0 ]; then
-            echo "::error::$STALE_COUNT tenant(s) returned a stale SHA. ssm_status=Success was misleading — see job summary."
-            exit 1
-          fi
-
-          echo "::notice::Tenant fleet redeploy complete — all reachable tenants on ${EXPECTED_SHA:0:7} (${UNREACHABLE_COUNT} unreachable, soft-warned)."
@@ -1,362 +0,0 @@
-name: redeploy-tenants-on-staging
-
-# Auto-refresh staging tenant EC2s after every staging-branch merge.
-#
-# Mirror of redeploy-tenants-on-main.yml, with the staging-CP host and
-# the :staging-latest tag. Sister workflow exists for prod (rolls
-# :latest after canary-verify). Both share the same shape — just
-# different CP_URL + target_tag + admin token secret.
-#
-# Why this workflow exists: publish-workspace-server-image now builds
-# on every staging-branch push (PR #2335), pushing
-# platform-tenant:staging-latest to GHCR. Existing tenants pulled
-# their image once at boot and never re-pull, so the new image just
-# sits unused until the tenant is reprovisioned.
-#
-# This workflow closes the gap by calling staging-CP's
-# /cp/admin/tenants/redeploy-fleet, which performs a canary-first,
-# batched, health-gated SSM redeploy across every live staging tenant.
-# Same endpoint shape as prod CP — only the host differs.
-#
-# Runtime ordering:
-#   1. publish-workspace-server-image completes on staging branch →
-#      new :staging-latest in GHCR.
-#   2. This workflow fires via workflow_run, waits 30s for GHCR's CDN
-#      to propagate the new tag.
-#   3. Calls redeploy-fleet with no canary (staging IS canary; we don't
-#      need a sub-canary inside it). Soak still applies to the first
-#      tenant in case of bad-deploy detection.
-#   4. Any failure aborts the rollout and leaves older tenants on the
-#      prior image — safer default than half-and-half state.
-#
-# Rollback path: re-run with workflow_dispatch + target_tag=staging-<sha>
-# of a known-good build.
-
-on:
-  workflow_run:
-    workflows: ['publish-workspace-server-image']
-    types: [completed]
-    branches: [main]
-  workflow_dispatch:
-    inputs:
-      target_tag:
-        description: 'Tenant image tag to deploy (e.g. "staging-latest" or "staging-a59f1a6c"). Defaults to staging-latest when empty.'
-        required: false
-        type: string
-        default: 'staging-latest'
-      canary_slug:
-        description: 'Tenant slug to deploy first + soak (empty = skip canary, fan out immediately). Default empty for staging since staging itself is the canary.'
-        required: false
-        type: string
-        default: ''
-      soak_seconds:
-        description: 'Seconds to wait after canary before fanning out. Only meaningful if canary_slug is set.'
-        required: false
-        type: string
-        default: '60'
-      batch_size:
-        description: 'How many tenants SSM redeploys in parallel per batch.'
-        required: false
-        type: string
-        default: '3'
-      dry_run:
-        description: 'Plan only — do not actually redeploy.'
-        required: false
-        type: boolean
-        default: false
-
-permissions:
-  contents: read
-  # No write scopes needed — the workflow hits an external CP endpoint,
-  # not the GitHub API.
-
-# Serialize per-branch so two rapid staging pushes' redeploys don't
-# overlap and cause confusing per-tenant SSM state. cancel-in-progress
-# is false because aborting a half-rolled-out fleet leaves tenants
-# stuck on whatever image they happened to be on when cancelled.
-concurrency:
-  group: redeploy-tenants-on-staging
-  cancel-in-progress: false
-
-jobs:
-  redeploy:
-    # Skip the auto-trigger if publish-workspace-server-image didn't
-    # actually succeed. workflow_run fires on any completion state; we
-    # don't want to redeploy against a half-built image.
-    if: |
-      github.event_name == 'workflow_dispatch' ||
-      (github.event_name == 'workflow_run' && github.event.workflow_run.conclusion == 'success')
-    runs-on: ubuntu-latest
-    timeout-minutes: 25
-    steps:
-      - name: Wait for GHCR tag propagation
-        # GHCR's edge cache takes ~15-30s to consistently serve the new
-        # :staging-latest manifest after the registry accepts the push.
-        # Same rationale as redeploy-tenants-on-main.yml.
-        run: sleep 30
-
-      - name: Call staging-CP redeploy-fleet
-        # CP_STAGING_ADMIN_API_TOKEN must be set as a repo/org secret
-        # on molecule-ai/molecule-core, matching staging-CP's
-        # CP_ADMIN_API_TOKEN env var (visible in Railway controlplane
-        # / staging environment). Stored separately from the prod
-        # CP_ADMIN_API_TOKEN so a leak of one doesn't auth the other.
-        env:
-          CP_URL: ${{ vars.STAGING_CP_URL || 'https://staging-api.moleculesai.app' }}
-          CP_STAGING_ADMIN_API_TOKEN: ${{ secrets.CP_STAGING_ADMIN_API_TOKEN }}
-          TARGET_TAG: ${{ inputs.target_tag || 'staging-latest' }}
-          CANARY_SLUG: ${{ inputs.canary_slug || '' }}
-          SOAK_SECONDS: ${{ inputs.soak_seconds || '60' }}
-          BATCH_SIZE: ${{ inputs.batch_size || '3' }}
-          DRY_RUN: ${{ inputs.dry_run || false }}
-        run: |
-          set -euo pipefail
-
-          # Schedule-vs-dispatch hardening (mirrors sweep-cf-orphans
-          # and sweep-cf-tunnels): hard-fail on auto-trigger when the
-          # secret is missing so a misconfigured-repo doesn't silently
-          # serve stale staging tenants. Soft-skip on operator dispatch.
-          if [ -z "${CP_STAGING_ADMIN_API_TOKEN:-}" ]; then
-            if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-              echo "::warning::CP_STAGING_ADMIN_API_TOKEN secret not set — skipping redeploy"
-              echo "::warning::Set CP_STAGING_ADMIN_API_TOKEN in repo secrets to enable auto-redeploy."
-              echo "::notice::Pull the value from staging-CP's CP_ADMIN_API_TOKEN env in Railway."
-              exit 0
-            fi
-            echo "::error::staging redeploy cannot run — CP_STAGING_ADMIN_API_TOKEN secret missing"
-            echo "::error::set it at Settings → Secrets and Variables → Actions; pull from staging-CP's CP_ADMIN_API_TOKEN env in Railway."
-            exit 1
-          fi
-
-          BODY=$(jq -nc \
-            --arg tag "$TARGET_TAG" \
-            --arg canary "$CANARY_SLUG" \
-            --argjson soak "$SOAK_SECONDS" \
-            --argjson batch "$BATCH_SIZE" \
-            --argjson dry "$DRY_RUN" \
-            '{
-              target_tag: $tag,
-              canary_slug: $canary,
-              soak_seconds: $soak,
-              batch_size: $batch,
-              dry_run: $dry
-            }')
-
-          echo "POST $CP_URL/cp/admin/tenants/redeploy-fleet"
-          echo "  body: $BODY"
-
-          HTTP_RESPONSE=$(mktemp)
-          HTTP_CODE_FILE=$(mktemp)
-          # Route -w into its own tempfile so curl's exit code (e.g. 56
-          # on connection-reset) can't pollute the captured stdout. The
-          # previous inline-substitution shape produced "000000" on
-          # connection reset — caught on main variant 2026-05-04
-          # redeploying sha 2b862f6. Same fix shape as the synth-E2E
-          # §9c gate (PR #2797). See lint-curl-status-capture.yml for
-          # the CI gate that pins this fix shape.
-          set +e
-          curl -sS -o "$HTTP_RESPONSE" -w '%{http_code}' \
-            -m 1200 \
-            -H "Authorization: Bearer $CP_STAGING_ADMIN_API_TOKEN" \
-            -H "Content-Type: application/json" \
-            -X POST "$CP_URL/cp/admin/tenants/redeploy-fleet" \
-            -d "$BODY" >"$HTTP_CODE_FILE"
-          set -e
-          # Stderr from curl (-sS shows dial errors etc.) goes to the
-          # runner log so operators can see WHY a connection failed.
-          HTTP_CODE=$(cat "$HTTP_CODE_FILE" 2>/dev/null || echo "000")
-          [ -z "$HTTP_CODE" ] && HTTP_CODE="000"
-
-          echo "HTTP $HTTP_CODE"
-          cat "$HTTP_RESPONSE" | jq . || cat "$HTTP_RESPONSE"
-
-          {
-            echo "## Staging tenant redeploy fleet"
-            echo ""
-            echo "**Target tag:** \`$TARGET_TAG\`"
-            echo "**Canary:** \`${CANARY_SLUG:-(none — staging is itself the canary)}\` (soak ${SOAK_SECONDS}s)"
-            echo "**Batch size:** $BATCH_SIZE"
-            echo "**Dry run:** $DRY_RUN"
-            echo "**HTTP:** $HTTP_CODE"
-            echo ""
-            echo "### Per-tenant result"
-            echo ""
-            echo '| Slug | Phase | SSM Status | Exit | Healthz | Error |'
-            echo '|------|-------|------------|------|---------|-------|'
-            jq -r '.results[]? | "| \(.slug) | \(.phase) | \(.ssm_status // "-") | \(.ssm_exit_code) | \(.healthz_ok) | \(.error // "-") |"' "$HTTP_RESPONSE" || true
-          } >> "$GITHUB_STEP_SUMMARY"
-
-          # Distinguish "real fleet failure" from "E2E teardown race".
-          #
-          # CP returns HTTP 500 + ok=false whenever ANY tenant in the
-          # fleet failed SSM or healthz. In practice the recurring source
-          # of these is ephemeral test tenants being torn down by their
-          # parent E2E run mid-redeploy: the EC2 dies → SSM exit=2 or
-          # healthz timeout → CP marks the fleet failed → this workflow
-          # goes red even though every operator-facing tenant rolled fine.
-          #
-          # Ephemeral slug prefixes (kept in sync with sweep-stale-e2e-orgs.yml
-          # — see that file for the source-of-truth list and rationale):
-          #   - e2e-*       — canvas/saas/ext E2E suites
-          #   - rt-e2e-*    — runtime-test harness fixtures (RFC #2251)
-          # Long-lived prefixes that are NOT ephemeral and MUST hard-fail:
-          # demo-prep, dryrun-*, dryrun2-*, plus all human tenant slugs.
-          #
-          # Filter: if HTTP=500/ok=false AND every failed slug matches an
-          # ephemeral prefix, treat as soft-warn and let the verify step
-          # downstream handle unreachable-vs-stale (#2402). Any non-ephemeral
-          # failure or a non-500 HTTP response remains a hard failure.
-          OK=$(jq -r '.ok // "false"' "$HTTP_RESPONSE")
-          FAILED_SLUGS=$(jq -r '
-            .results[]?
-            | select((.healthz_ok != true) or (.ssm_status != "Success"))
-            | .slug' "$HTTP_RESPONSE" 2>/dev/null || true)
-          EPHEMERAL_PREFIX_RE='^(e2e-|rt-e2e-)'
-          NON_EPHEMERAL_FAILED=$(printf '%s\n' "$FAILED_SLUGS" | grep -v '^$' | grep -Ev "$EPHEMERAL_PREFIX_RE" || true)
-
-          if [ "$HTTP_CODE" = "200" ] && [ "$OK" = "true" ]; then
-            : # happy path — fall through to verification
-          elif [ "$HTTP_CODE" = "500" ] && [ -z "$NON_EPHEMERAL_FAILED" ] && [ -n "$FAILED_SLUGS" ]; then
-            COUNT=$(printf '%s\n' "$FAILED_SLUGS" | grep -Ec "$EPHEMERAL_PREFIX_RE" || true)
-            echo "::warning::redeploy-fleet returned HTTP 500 but every failed tenant ($COUNT) is ephemeral (e2e-*/rt-e2e-*) — treating as teardown race, soft-warning."
-            printf '%s\n' "$FAILED_SLUGS" | sed 's/^/::warning::  failed: /'
-          elif [ "$HTTP_CODE" != "200" ]; then
-            echo "::error::redeploy-fleet returned HTTP $HTTP_CODE"
-            if [ -n "$NON_EPHEMERAL_FAILED" ]; then
-              echo "::error::non-ephemeral tenant(s) failed:"
-              printf '%s\n' "$NON_EPHEMERAL_FAILED" | sed 's/^/::error::  /'
-            fi
-            exit 1
-          else
-            # HTTP=200 but ok=false (shouldn't happen with current CP
-            # but keep the gate for completeness).
-            echo "::error::redeploy-fleet reported ok=false (see summary for which tenant halted the rollout)"
-            exit 1
-          fi
-          echo "::notice::Staging tenant fleet redeploy reported ssm_status=Success — verifying actual image roll on each tenant..."
-
-          cp "$HTTP_RESPONSE" "$RUNNER_TEMP/redeploy-response.json"
-
-      - name: Verify each staging tenant /buildinfo matches published SHA
-        # Mirror of the verify step in redeploy-tenants-on-main.yml — see
-        # there for the rationale (#2395 root fix). Staging has the same
-        # ssm_status-success-but-stale-image hazard and benefits from the
-        # same gate. Diff: TENANT_DOMAIN includes the `staging.` infix.
-        env:
-          EXPECTED_SHA: ${{ github.event.workflow_run.head_sha || github.sha }}
-          TARGET_TAG: ${{ inputs.target_tag || 'staging-latest' }}
-          TENANT_DOMAIN: 'staging.moleculesai.app'
-        run: |
-          set -euo pipefail
-
-          # staging-latest is the staging-side moving tag; treat it the
-          # same way main treats `latest`. Operator-pinned SHAs skip
-          # verification (see main variant for why).
-          if [ "$TARGET_TAG" != "staging-latest" ] && [ "$TARGET_TAG" != "latest" ] && [ "$TARGET_TAG" != "$EXPECTED_SHA" ]; then
-            echo "::notice::target_tag=$TARGET_TAG (operator-pinned) — skipping per-tenant SHA verification."
-            exit 0
-          fi
-
-          RESP="$RUNNER_TEMP/redeploy-response.json"
-          if [ ! -s "$RESP" ]; then
-            echo "::error::redeploy-response.json missing or empty"
-            exit 1
-          fi
-
-          mapfile -t SLUGS < <(jq -r '.results[]? | select(.healthz_ok == true) | .slug' "$RESP")
-          if [ ${#SLUGS[@]} -eq 0 ]; then
-            echo "::warning::No staging tenants reported healthz_ok — nothing to verify"
-            exit 0
-          fi
-
-          echo "Verifying ${#SLUGS[@]} staging tenant(s) against EXPECTED_SHA=${EXPECTED_SHA:0:7}..."
-
-          # Two distinct failure modes here:
-          #   STALE_COUNT      — tenant returned a SHA that doesn't match. THIS is
-          #                      the #2395 bug class: tenant up + serving old code.
-          #                      Always hard-fail the workflow.
-          #   UNREACHABLE_COUNT — tenant didn't respond. Almost always a benign
-          #                      teardown race: redeploy-fleet snapshot says
-          #                      healthz_ok=true, then the E2E suite tears the
-          #                      ephemeral tenant down before this step runs (the
-          #                      e2e-* fixtures churn 5-10/hour on staging). Soft-
-          #                      warn so we don't block staging→main on cleanup.
-          #                      Real "tenant up but unreachable" is caught by CP's
-          #                      own healthz monitor + the post-redeploy alert; we
-          #                      don't need to double-count it here.
-          STALE_COUNT=0
-          UNREACHABLE_COUNT=0
-          STALE_LINES=()
-          UNREACHABLE_LINES=()
-          for slug in "${SLUGS[@]}"; do
-            URL="https://${slug}.${TENANT_DOMAIN}/buildinfo"
-            BODY=$(curl -sS --max-time 30 --retry 3 --retry-delay 5 --retry-connrefused "$URL" || true)
-            ACTUAL_SHA=$(echo "$BODY" | jq -r '.git_sha // ""' 2>/dev/null || echo "")
-            if [ -z "$ACTUAL_SHA" ]; then
-              UNREACHABLE_COUNT=$((UNREACHABLE_COUNT + 1))
-              UNREACHABLE_LINES+=("| $slug | (no /buildinfo response) | ${EXPECTED_SHA:0:7} | ⚠ unreachable (likely teardown race) |")
-              continue
-            fi
-            if [ "$ACTUAL_SHA" = "$EXPECTED_SHA" ]; then
-              echo "  $slug: ${ACTUAL_SHA:0:7} ✓"
-            else
-              STALE_COUNT=$((STALE_COUNT + 1))
-              STALE_LINES+=("| $slug | ${ACTUAL_SHA:0:7} | ${EXPECTED_SHA:0:7} | ❌ stale |")
-            fi
-          done
-
-          {
-            echo ""
-            echo "### Per-tenant /buildinfo verification (staging)"
-            echo ""
-            echo "Expected SHA: \`${EXPECTED_SHA:0:7}\`"
-            echo ""
-            if [ $STALE_COUNT -gt 0 ]; then
-              echo "**${STALE_COUNT} STALE tenant(s) — these did NOT pick up the new image despite ssm_status=Success:**"
-              echo ""
-              echo "| Slug | Actual /buildinfo SHA | Expected | Status |"
-              echo "|------|----------------------|----------|--------|"
-              for line in "${STALE_LINES[@]}"; do echo "$line"; done
-              echo ""
-            fi
-            if [ $UNREACHABLE_COUNT -gt 0 ]; then
-              echo "**${UNREACHABLE_COUNT} unreachable tenant(s) — likely E2E teardown race (soft-warn, not failing):**"
-              echo ""
-              echo "| Slug | Actual /buildinfo SHA | Expected | Status |"
-              echo "|------|----------------------|----------|--------|"
-              for line in "${UNREACHABLE_LINES[@]}"; do echo "$line"; done
-              echo ""
-            fi
-            if [ $STALE_COUNT -eq 0 ] && [ $UNREACHABLE_COUNT -eq 0 ]; then
-              echo "All ${#SLUGS[@]} staging tenants returned matching SHA. ✓"
-            fi
-          } >> "$GITHUB_STEP_SUMMARY"
-
-          if [ $UNREACHABLE_COUNT -gt 0 ]; then
-            echo "::warning::$UNREACHABLE_COUNT staging tenant(s) unreachable post-redeploy. Likely benign teardown race — CP healthz monitor catches real outages."
-          fi
-
-          # Belt-and-suspenders sanity floor: if MORE than half the fleet is
-          # unreachable AND the fleet is large enough that "half down" is
-          # statistically meaningful, this is a real outage (e.g. new image
-          # crashes on startup), not a teardown race. Hard-fail.
-          #
-          # Floor only applies when TOTAL_VERIFIED >= 4 — below that, the
-          # canary-verify step is the actual gate for "all tenants down"
-          # detection (it runs against the canary first and aborts the
-          # rollout if the canary fails to come up). Without the >=4 gate,
-          # a 1-tenant fleet (e.g. a single ephemeral e2e-* tenant on a
-          # quiet staging push) would re-flake on the exact teardown-race
-          # condition #2402 fixed: 1 of 1 unreachable = 100% > 50% → fail.
-          TOTAL_VERIFIED=${#SLUGS[@]}
-          if [ $TOTAL_VERIFIED -ge 4 ] && [ $UNREACHABLE_COUNT -gt $((TOTAL_VERIFIED / 2)) ]; then
-            echo "::error::$UNREACHABLE_COUNT of $TOTAL_VERIFIED staging tenant(s) unreachable — exceeds 50% threshold on a fleet large enough that this signals a real outage, not teardown race."
-            exit 1
-          fi
-
-          if [ $STALE_COUNT -gt 0 ]; then
-            echo "::error::$STALE_COUNT staging tenant(s) returned a stale SHA. ssm_status=Success was misleading — see job summary."
-            exit 1
-          fi
-
-          echo "::notice::Staging tenant fleet redeploy complete — all reachable tenants on ${EXPECTED_SHA:0:7} (${UNREACHABLE_COUNT} unreachable, soft-warned)."
@@ -1,91 +0,0 @@
-name: Runtime Pin Compatibility
-
-# CI gate that prevents the 5-hour staging outage from 2026-04-24 from
-# recurring (controlplane#253). The original failure mode:
-#   1. molecule-ai-workspace-runtime 0.1.13 declared `a2a-sdk<1.0` in its
-#      requires_dist metadata (incorrect — it actually imports
-#      a2a.server.routes which only exists in a2a-sdk 1.0+)
-#   2. `pip install molecule-ai-workspace-runtime` resolved cleanly
-#   3. `from molecule_runtime.main import main_sync` raised ImportError
-#   4. Every tenant workspace crashed; the canary tenant caught it but
-#      only after 5 hours of degraded staging
-#
-# This workflow installs the CURRENTLY PUBLISHED runtime from PyPI on
-# top of `workspace/requirements.txt` and smoke-imports. Catches:
-#   - Upstream PyPI yanks
-#   - Bad re-releases of molecule-ai-workspace-runtime
-#   - Already-shipped wheels that stop importing because a transitive
-#     dep moved underneath
-#
-# This is the "PyPI artifact health" half of pin compatibility. The
-# companion workflow `runtime-prbuild-compat.yml` covers the
-# "PR-introduced breakage" half by building the wheel from THIS PR's
-# workspace/ source. Splitting the two means each gets a narrow
-# `paths:` filter — the pypi-latest job no longer fires on doc-only
-# workspace/ edits whose content can't change what's currently on PyPI.
-
-on:
-  push:
-    branches: [main, staging]
-    paths:
-      # Narrow filter: pypi-latest is sensitive only to changes that
-      # affect what we're INSTALLING (requirements.txt) or WHAT THE
-      # CHECK ITSELF DOES (this workflow file). Edits to workspace/
-      # source code don't change what's on PyPI right now, so they
-      # don't change this gate's verdict.
-      - 'workspace/requirements.txt'
-      - '.github/workflows/runtime-pin-compat.yml'
-  pull_request:
-    branches: [main, staging]
-    paths:
-      - 'workspace/requirements.txt'
-      - '.github/workflows/runtime-pin-compat.yml'
-  # Daily catch for upstream PyPI publishes that break the pin combo
-  # without any change in our repo (e.g. someone re-yanks an a2a-sdk
-  # release or molecule-ai-workspace-runtime publishes a bad bump).
-  schedule:
-    - cron: '0 13 * * *'  # 06:00 PT
-  workflow_dispatch:
-  # Required-check support: when this becomes a branch-protection gate,
-  # merge_group runs let the queue green-check this in addition to PRs.
-  merge_group:
-    types: [checks_requested]
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  pypi-latest-install:
-    name: PyPI-latest install + import smoke
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
-        with:
-          python-version: '3.11'
-          cache: pip
-          cache-dependency-path: workspace/requirements.txt
-      - name: Install runtime + workspace requirements
-        # Install order is load-bearing: install the runtime FIRST so pip
-        # honors whatever a2a-sdk constraint the runtime metadata declares
-        # (this is the surface that broke in 2026-04-24 — runtime declared
-        # `a2a-sdk<1.0` but actually needed >=1.0). The follow-up install
-        # of workspace/requirements.txt then upgrades a2a-sdk to the
-        # constraint our runtime image actually pins. The import smoke
-        # below verifies the upgraded combination is consistent.
-        run: |
-          python -m venv /tmp/venv
-          /tmp/venv/bin/pip install --upgrade pip
-          /tmp/venv/bin/pip install molecule-ai-workspace-runtime
-          /tmp/venv/bin/pip install -r workspace/requirements.txt
-          /tmp/venv/bin/pip show molecule-ai-workspace-runtime a2a-sdk \
-            | grep -E '^(Name|Version):'
-      - name: Smoke import — fail if metadata declares deps that don't satisfy real imports
-        # WORKSPACE_ID is validated at import time by platform_auth.py — EC2
-        # user-data sets it from the cloud-init template; set a placeholder
-        # here so the import smoke doesn't trip on the env-var guard.
-        env:
-          WORKSPACE_ID: 00000000-0000-0000-0000-000000000001
-        run: |
-          /tmp/venv/bin/python -c "from molecule_runtime.main import main_sync; print('runtime imports OK')"
@@ -1,152 +0,0 @@
-name: Runtime PR-Built Compatibility
-
-# Companion to `runtime-pin-compat.yml`. That workflow tests what's
-# CURRENTLY PUBLISHED on PyPI; this workflow tests what WOULD BE
-# PUBLISHED if THIS PR merges.
-#
-# Why two workflows: the chicken-and-egg #128 fix added a "PR-built
-# wheel" job to the original runtime-pin-compat.yml, but both jobs
-# shared a `paths:` filter that was the union of their needs
-# (`workspace/**`). That meant the PyPI-latest job ran on every doc
-# edit even though the upstream PyPI artifact can't change with our
-# workspace/ source. Splitting the two means each gets a narrow
-# `paths:` filter that matches the inputs it actually depends on.
-#
-# Catches the failure mode where a PR adds an import requiring a newer
-# SDK than `workspace/requirements.txt` pins:
-#   1. Pip resolves the existing PyPI wheel + the old SDK pin → smoke
-#      passes (it imports the OLD main.py from the wheel, not the PR's
-#      new main.py).
-#   2. Merge → publish-runtime.yml ships a wheel WITH the new import.
-#   3. Tenant images redeploy → all crash on first boot with
-#      ImportError.
-#
-# By building from the PR's source and smoke-importing THAT wheel, we
-# fail at PR-time instead of after publish.
-#
-# Required-check shape (2026-05-01): the workflow runs on EVERY push +
-# PR + merge_group event with no top-level `paths:` filter, then uses a
-# detect-changes job + per-step `if:` gates inside ONE always-running
-# job named `PR-built wheel + import smoke`. PRs that don't touch
-# wheel-relevant paths get a no-op SUCCESS check run, satisfying branch
-# protection without re-running the heavy build. Same pattern as
-# e2e-api.yml — see its comment for the full rationale + the 2026-04-29
-# PR #2264 incident that motivated the always-run-with-if-gates shape.
-
-on:
-  push:
-    branches: [main, staging]
-  pull_request:
-    branches: [main, staging]
-  workflow_dispatch:
-  merge_group:
-    types: [checks_requested]
-
-concurrency:
-  # Include event_name so a PR sync (event=pull_request) and the
-  # subsequent staging push (event=push) on the SAME merge SHA don't
-  # collide in one group. Without event_name, both runs hashed to
-  # the same key and cancel-in-progress=true cancelled whichever
-  # arrived second — usually the push run, which staging branch-
-  # protection then sees as a CANCELLED required check and refuses
-  # to mark merged. Caught 2026-05-05 across PR #2869's runs (run
-  # ids 25371863455 / 25371811486 / 25371078157 / 25370403142 — every
-  # staging push run cancelled, every matching PR run green).
-  #
-  # Per memory `feedback_concurrency_group_per_sha.md` — same drift
-  # class that broke auto-promote-staging on 2026-04-28. Pin invariant:
-  # event_name + sha is the minimum unique key for these workflows.
-  group: ${{ github.workflow }}-${{ github.event_name }}-${{ github.event.pull_request.head.sha || github.sha }}
-  cancel-in-progress: true
-
-jobs:
-  detect-changes:
-    runs-on: ubuntu-latest
-    outputs:
-      wheel: ${{ steps.decide.outputs.wheel }}
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: dorny/paths-filter@fbd0ab8f3e69293af611ebaee6363fc25e6d187d # v4.0.1
-        id: filter
-        with:
-          filters: |
-            wheel:
-              - 'workspace/**'
-              - 'scripts/build_runtime_package.py'
-              - 'scripts/wheel_smoke.py'
-              - '.github/workflows/runtime-prbuild-compat.yml'
-      - id: decide
-        # Always run real work for manual dispatch + merge_group — no
-        # diff-against-base in those contexts, and the gate exists to
-        # validate the to-be-merged state regardless of which paths it
-        # touched (paths-filter would default to "no changes" which is
-        # the wrong answer when the queue is composing many PRs).
-        run: |
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ] || [ "${{ github.event_name }}" = "merge_group" ]; then
-            echo "wheel=true" >> "$GITHUB_OUTPUT"
-          else
-            echo "wheel=${{ steps.filter.outputs.wheel }}" >> "$GITHUB_OUTPUT"
-          fi
-
-  # ONE job (no job-level `if:`) that always runs and reports under the
-  # required-check name `PR-built wheel + import smoke`. Real work is
-  # gated per-step on `needs.detect-changes.outputs.wheel`. Same shape
-  # as e2e-api.yml's e2e-api job — see its comment block for the full
-  # rationale (SKIPPED check runs block branch protection even with
-  # SUCCESS siblings; collapsing to one always-run job emits exactly
-  # one SUCCESS check run).
-  local-build-install:
-    needs: detect-changes
-    name: PR-built wheel + import smoke
-    runs-on: ubuntu-latest
-    steps:
-      - name: No-op pass (paths filter excluded this commit)
-        if: needs.detect-changes.outputs.wheel != 'true'
-        run: |
-          echo "No workspace/ / scripts/{build_runtime_package,wheel_smoke}.py / workflow changes — wheel gate satisfied without rebuilding."
-          echo "::notice::PR-built wheel + import smoke no-op pass (paths filter excluded this commit)."
-      - if: needs.detect-changes.outputs.wheel == 'true'
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.detect-changes.outputs.wheel == 'true'
-        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
-        with:
-          python-version: '3.11'
-          cache: pip
-          cache-dependency-path: workspace/requirements.txt
-      - name: Install build tooling
-        if: needs.detect-changes.outputs.wheel == 'true'
-        run: pip install build
-      - name: Build wheel from PR source (mirrors publish-runtime.yml)
-        if: needs.detect-changes.outputs.wheel == 'true'
-        # Use a fixed test version so the wheel filename is predictable.
-        # Doesn't reach PyPI — this build is local-only for the smoke.
-        # Use the SAME build script with the SAME args as
-        # publish-runtime.yml's build step. The temp dir path differs
-        # (`/tmp/runtime-build` here vs `${{ runner.temp }}/runtime-build`
-        # in publish-runtime.yml — they coincide on ubuntu-latest but
-        # the call sites are not byte-identical). The smoke import is
-        # also intentionally narrower than publish's: this gate exists
-        # to catch SDK-version-import drift specifically; full invariant
-        # coverage lives in publish-runtime.yml's own pre-PyPI smoke.
-        run: |
-          python scripts/build_runtime_package.py \
-            --version "0.0.0.dev0+pin-compat" \
-            --out /tmp/runtime-build
-          cd /tmp/runtime-build && python -m build
-      - name: Install built wheel + workspace requirements
-        if: needs.detect-changes.outputs.wheel == 'true'
-        run: |
-          python -m venv /tmp/venv-built
-          /tmp/venv-built/bin/pip install --upgrade pip
-          /tmp/venv-built/bin/pip install /tmp/runtime-build/dist/*.whl
-          /tmp/venv-built/bin/pip install -r workspace/requirements.txt
-          /tmp/venv-built/bin/pip show molecule-ai-workspace-runtime a2a-sdk \
-            | grep -E '^(Name|Version):'
-      - name: Smoke import the PR-built wheel
-        if: needs.detect-changes.outputs.wheel == 'true'
-        # Same script publish-runtime.yml runs against the to-be-PyPI wheel.
-        # Closes the PR-time vs publish-time gap: a PR adding a new SDK
-        # call-shape no longer passes here (narrow `import main_sync`) only
-        # to fail post-merge in publish-runtime's broader smoke.
-        run: |
-          /tmp/venv-built/bin/python "$GITHUB_WORKSPACE/scripts/wheel_smoke.py"
@@ -1,58 +0,0 @@
-name: SECRET_PATTERNS drift lint
-
-# Detects when the canonical SECRET_PATTERNS array in
-# .github/workflows/secret-scan.yml diverges from known consumer
-# mirrors (workspace-runtime's bundled pre-commit hook today; more
-# can be added as the consumer set grows).
-#
-# Why this exists: every side that scans for credentials has its own
-# copy of the pattern list. They drift — most recently the runtime
-# hook lagged the canonical by one pattern (sk-cp- / MiniMax F1088),
-# so a developer's local pre-commit would let a sk-cp- token through
-# while the org-wide CI scan would refuse it. The cost of that drift
-# is dev confusion + delayed feedback; the fix is automated detection.
-#
-# Triggers:
-#   - schedule: daily 05:00 UTC. Catches drift introduced by edits
-#     to a consumer copy that didn't update canonical here.
-#   - push to main/staging where the canonical or this lint changed:
-#     catches the inverse — canonical updated but consumers not yet
-#     bumped. The lint will fail the push; that's intentional, the
-#     person editing canonical is the right person to also update
-#     the consumer.
-#   - workflow_dispatch: ad-hoc operator runs.
-
-on:
-  schedule:
-    # 05:00 UTC = 22:00 PT / 01:00 ET. Quiet hours so a failure
-    # email lands when humans are starting their day, not
-    # interrupting it.
-    - cron: "0 5 * * *"
-  push:
-    branches: [main, staging]
-    paths:
-      - ".github/workflows/secret-scan.yml"
-      - ".github/workflows/secret-pattern-drift.yml"
-      - ".github/scripts/lint_secret_pattern_drift.py"
-      - ".githooks/pre-commit"
-  workflow_dispatch:
-
-# GITHUB_TOKEN scoped to read-only. The lint only does git checkout
-# + HTTPS GETs to public consumer files; no writes to anything.
-permissions:
-  contents: read
-
-jobs:
-  lint:
-    name: Detect SECRET_PATTERNS drift
-    runs-on: ubuntu-latest
-    timeout-minutes: 5
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
-        with:
-          python-version: "3.11"
-
-      - name: Run drift lint
-        run: python3 .github/scripts/lint_secret_pattern_drift.py
@@ -1,129 +0,0 @@
-name: Sweep stale AWS Secrets Manager secrets
-
-# Janitor for per-tenant AWS Secrets Manager secrets
-# (`molecule/tenant/<org_id>/bootstrap`) whose backing tenant no
-# longer exists. Parallel-shape to sweep-cf-tunnels.yml and
-# sweep-cf-orphans.yml — different cloud, same justification.
-#
-# Why this exists separately from a long-term reconciler integration:
-#   - molecule-controlplane's tenant_resources audit table (mig 024)
-#     currently tracks four resource kinds: CloudflareTunnel,
-#     CloudflareDNS, EC2Instance, SecurityGroup. SecretsManager is
-#     not in the list, so the existing reconciler doesn't catch
-#     orphan secrets.
-#   - At ~$0.40/secret/month the cost grew to ~$19/month before this
-#     sweeper was written, indicating ~45+ orphan secrets from
-#     crashed provisions and incomplete deprovision flows.
-#   - The proper fix (KindSecretsManagerSecret + recorder hook +
-#     reconciler enumerator) is filed as a separate controlplane
-#     issue. This sweeper is the immediate cost-relief stopgap.
-#
-# IAM principal: AWS_JANITOR_ACCESS_KEY_ID / AWS_JANITOR_SECRET_ACCESS_KEY.
-# This is a DEDICATED principal — the production `molecule-cp` IAM
-# user lacks `secretsmanager:ListSecrets` (it only has
-# Get/Create/Update/Delete on specific resources, scoped to its
-# operational needs). The janitor needs ListSecrets across the
-# `molecule/tenant/*` prefix, which warrants a separate principal so
-# we don't broaden the prod-CP policy.
-#
-# Safety: the script's MAX_DELETE_PCT gate (default 50%, mirroring
-# sweep-cf-orphans.yml — tenant secrets are durable by design, unlike
-# the mostly-orphan tunnels) refuses to nuke past the threshold.
-
-on:
-  schedule:
-    # Hourly at :30 — offsets from sweep-cf-orphans (:15) and
-    # sweep-cf-tunnels (:45) so the three janitors don't burst the
-    # CP admin endpoints at the same minute.
-    - cron: '30 * * * *'
-  workflow_dispatch:
-    inputs:
-      dry_run:
-        description: "Dry run only — list what would be deleted, no deletion"
-        required: false
-        type: boolean
-        default: true
-      max_delete_pct:
-        description: "Override safety gate (default 50, set higher only for major cleanup)"
-        required: false
-        default: "50"
-      grace_hours:
-        description: "Skip secrets created within this many hours (default 24)"
-        required: false
-        default: "24"
-
-# Don't let two sweeps race the same AWS account.
-concurrency:
-  group: sweep-aws-secrets
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  sweep:
-    name: Sweep AWS Secrets Manager
-    runs-on: ubuntu-latest
-    # 30 min cap, mirroring the other janitors. AWS DeleteSecret is
-    # fast (~0.3s/call) so even a 100+ backlog drains in seconds
-    # under the 8-way xargs parallelism, but the cap is set generously
-    # to leave headroom for any actual API hang.
-    timeout-minutes: 30
-    env:
-      AWS_REGION: ${{ secrets.AWS_REGION || 'us-east-1' }}
-      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_JANITOR_ACCESS_KEY_ID }}
-      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_JANITOR_SECRET_ACCESS_KEY }}
-      CP_PROD_ADMIN_TOKEN: ${{ secrets.CP_PROD_ADMIN_TOKEN }}
-      CP_STAGING_ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_TOKEN }}
-      MAX_DELETE_PCT: ${{ github.event.inputs.max_delete_pct || '50' }}
-      GRACE_HOURS: ${{ github.event.inputs.grace_hours || '24' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify required secrets present
-        id: verify
-        # Schedule-vs-dispatch behaviour split mirrors sweep-cf-orphans
-        # and sweep-cf-tunnels (hardened 2026-04-28). Same principle:
-        #   - schedule → exit 1 on missing secrets (red CI surfaces it)
-        #   - workflow_dispatch → exit 0 with warning (operator-driven,
-        #     they already accepted the repo state)
-        run: |
-          missing=()
-          for var in AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY CP_PROD_ADMIN_TOKEN CP_STAGING_ADMIN_TOKEN; do
-            if [ -z "${!var:-}" ]; then
-              missing+=("$var")
-            fi
-          done
-          if [ ${#missing[@]} -gt 0 ]; then
-            if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-              echo "::warning::skipping sweep — secrets not configured: ${missing[*]}"
-              echo "::warning::set them at Settings → Secrets and Variables → Actions, then rerun."
-              echo "::warning::AWS_JANITOR_* must belong to a principal with secretsmanager:ListSecrets and secretsmanager:DeleteSecret on molecule/tenant/* (the prod molecule-cp principal lacks ListSecrets)."
-              echo "skip=true" >> "$GITHUB_OUTPUT"
-              exit 0
-            fi
-            echo "::error::sweep cannot run — required secrets missing: ${missing[*]}"
-            echo "::error::set them at Settings → Secrets and Variables → Actions, or disable this workflow."
-            echo "::error::AWS_JANITOR_* must belong to a principal with secretsmanager:ListSecrets and secretsmanager:DeleteSecret on molecule/tenant/*."
-            exit 1
-          fi
-          echo "All required secrets present ✓"
-          echo "skip=false" >> "$GITHUB_OUTPUT"
-
-      - name: Run sweep
-        if: steps.verify.outputs.skip != 'true'
-        # Schedule-vs-dispatch dry-run asymmetry mirrors sweep-cf-tunnels:
-        #   - Scheduled: input empty → "false" → --execute (the whole
-        #     point of an hourly janitor).
-        #   - Manual workflow_dispatch: input default true → dry-run;
-        #     operator must flip it to actually delete.
-        run: |
-          set -euo pipefail
-          if [ "${{ github.event.inputs.dry_run || 'false' }}" = "true" ]; then
-            echo "Running in dry-run mode — no deletions"
-            bash scripts/ops/sweep-aws-secrets.sh
-          else
-            echo "Running with --execute — will delete identified orphans"
-            bash scripts/ops/sweep-aws-secrets.sh --execute
-          fi
@@ -1,146 +0,0 @@
-name: Sweep stale Cloudflare DNS records
-
-# Janitor for Cloudflare DNS records whose backing tenant/workspace no
-# longer exists. Without this loop, every short-lived E2E or canary
-# leaves a CF record on the moleculesai.app zone — the zone has a
-# 200-record quota (controlplane#239 hit it 2026-04-23+) and provisions
-# start failing with code 81045 once exhausted.
-#
-# Why a separate workflow vs sweep-stale-e2e-orgs.yml:
-#   - That workflow operates at the CP layer (DELETE /cp/admin/tenants/:slug
-#     drives the cascade). It assumes CP has the org row to drive the
-#     deprovision from. It doesn't catch records left behind when CP
-#     itself never knew about the tenant (canary scratch, manual ops
-#     experiments) or when the cascade's CF-delete branch failed.
-#   - sweep-cf-orphans.sh enumerates the CF zone directly and matches
-#     each record against live CP slugs + AWS EC2 names. It catches
-#     leaks the CP-driven sweep can't.
-#
-# Safety: the script's own MAX_DELETE_PCT gate refuses to nuke more
-# than 50% of records in a single run. If something has gone weird
-# (CP admin endpoint returns no orgs → every tenant looks orphan) the
-# gate halts before damage. Decision-function unit tests in
-# scripts/ops/test_sweep_cf_decide.py (#2027) cover the rule
-# classifier.
-
-on:
-  schedule:
-    # Hourly. Mirrors sweep-stale-e2e-orgs cadence so the two janitors
-    # converge on the same tick. CF API rate budget is generous (1200
-    # req/5min); a single sweep makes ~1 list + N deletes (N<=quota/2).
-    - cron: '15 * * * *'  # offset from sweep-stale-e2e-orgs (top of hour)
-  workflow_dispatch:
-    inputs:
-      dry_run:
-        description: "Dry run only — list what would be deleted, no deletion"
-        required: false
-        type: boolean
-        default: true
-      max_delete_pct:
-        description: "Override safety gate (default 50, set higher only for major cleanup)"
-        required: false
-        default: "50"
-  # No `merge_group:` trigger on purpose. This is a janitor — it doesn't
-  # need to gate merges, and including it as written before #2088 fired
-  # the full sweep job (or its secret-check) on every PR going through
-  # the merge queue, generating one red CI run per merge-queue eval. If
-  # this workflow is ever wired up as a required check, re-add
-  #   merge_group: { types: [checks_requested] }
-  # AND gate the sweep step with `if: github.event_name != 'merge_group'`
-  # so merge-queue evals report success without actually running.
-
-# Don't let two sweeps race the same zone. workflow_dispatch during a
-# scheduled run would otherwise issue duplicate DELETE calls.
-concurrency:
-  group: sweep-cf-orphans
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  sweep:
-    name: Sweep CF orphans
-    runs-on: ubuntu-latest
-    # 3 min surfaces hangs (CF API stall, AWS describe-instances stuck)
-    # within one cron interval instead of burning a full tick. Realistic
-    # worst case is ~2 min: 4 sequential curls + 1 aws + N×CF-DELETE
-    # each individually capped at 10s by the script's curl -m flag.
-    timeout-minutes: 3
-    env:
-      CF_API_TOKEN: ${{ secrets.CF_API_TOKEN }}
-      CF_ZONE_ID: ${{ secrets.CF_ZONE_ID }}
-      CP_PROD_ADMIN_TOKEN: ${{ secrets.CP_PROD_ADMIN_TOKEN }}
-      CP_STAGING_ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_TOKEN }}
-      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-      AWS_DEFAULT_REGION: us-east-2
-      MAX_DELETE_PCT: ${{ github.event.inputs.max_delete_pct || '50' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify required secrets present
-        id: verify
-        # Schedule-vs-dispatch behaviour split (hardened 2026-04-28
-        # after the silent-no-op incident below):
-        #
-        # The earlier soft-skip-on-schedule policy hid a real leak. All
-        # six secrets were unset on this repo for an unknown duration;
-        # every hourly run printed a yellow ::warning:: and exited 0,
-        # so the workflow registered as "passing" while doing nothing.
-        # CF orphans accumulated to 152/200 (~76% of the zone quota
-        # gone) before a manual `dig`-driven audit caught it. Anything
-        # that runs as a janitor and reports green while idle is
-        # indistinguishable from "the janitor is healthy" — so we now
-        # treat schedule (and any future workflow_run/push triggers)
-        # as a hard-fail when secrets are missing.
-        #
-        #   - schedule / workflow_run / push → exit 1 (red CI run
-        #     surfaces the misconfiguration the next tick)
-        #   - workflow_dispatch              → exit 0 with a warning
-        #     (an operator ran this ad-hoc; they already accepted the
-        #     state of the repo and want the workflow to short-circuit
-        #     so they can rerun after fixing the secret)
-        run: |
-          missing=()
-          for var in CF_API_TOKEN CF_ZONE_ID CP_PROD_ADMIN_TOKEN CP_STAGING_ADMIN_TOKEN AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY; do
-            if [ -z "${!var:-}" ]; then
-              missing+=("$var")
-            fi
-          done
-          if [ ${#missing[@]} -gt 0 ]; then
-            if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-              echo "::warning::skipping sweep — secrets not configured: ${missing[*]}"
-              echo "::warning::set them at Settings → Secrets and Variables → Actions, then rerun."
-              echo "skip=true" >> "$GITHUB_OUTPUT"
-              exit 0
-            fi
-            echo "::error::sweep cannot run — required secrets missing: ${missing[*]}"
-            echo "::error::set them at Settings → Secrets and Variables → Actions, or disable this workflow."
-            echo "::error::a silent skip masked an active CF DNS leak (152/200 zone records) caught only by a manual audit on 2026-04-28; this gate exists to make the gap visible."
-            exit 1
-          fi
-          echo "All required secrets present ✓"
-          echo "skip=false" >> "$GITHUB_OUTPUT"
-
-      - name: Run sweep
-        if: steps.verify.outputs.skip != 'true'
-        # Schedule-vs-dispatch dry-run asymmetry (intentional):
-        #   - Scheduled runs: github.event.inputs.dry_run is empty →
-        #     defaults to "false" below → script runs with --execute
-        #     (the whole point of an hourly janitor).
-        #   - Manual workflow_dispatch: input default is true (line 38)
-        #     so an ad-hoc operator-triggered run is dry-run by default;
-        #     they have to flip the toggle to actually delete.
-        # The script's MAX_DELETE_PCT gate (default 50%) is the second
-        # line of defense regardless of mode.
-        run: |
-          set -euo pipefail
-          if [ "${{ github.event.inputs.dry_run || 'false' }}" = "true" ]; then
-            echo "Running in dry-run mode — no deletions"
-            bash scripts/ops/sweep-cf-orphans.sh
-          else
-            echo "Running with --execute — will delete identified orphans"
-            bash scripts/ops/sweep-cf-orphans.sh --execute
-          fi
@@ -1,124 +0,0 @@
-name: Sweep stale Cloudflare Tunnels
-
-# Janitor for Cloudflare Tunnels whose backing tenant no longer
-# exists. Parallel-shape to sweep-cf-orphans.yml (which sweeps DNS
-# records); same justification, different CF resource.
-#
-# Why this exists separately from sweep-cf-orphans:
-#   - DNS records live on the zone (`/zones/<id>/dns_records`).
-#   - Tunnels live on the account (`/accounts/<id>/cfd_tunnel`).
-#   - Different CF API surface, different scopes; the existing CF
-#     token might not have `account:cloudflare_tunnel:edit`. Splitting
-#     the workflows keeps each one's secret-presence gate independent
-#     so neither silent-skips when the other's secret is missing.
-#   - Cleaner blast radius — operators can disable one without the
-#     other if a regression surfaces.
-#
-# Safety: the script's MAX_DELETE_PCT gate (default 90% — higher than
-# the DNS sweep's 50% because tenant-shaped tunnels are mostly
-# orphans by design) refuses to nuke past the threshold.
-
-on:
-  schedule:
-    # Hourly at :45 — offset from sweep-cf-orphans (:15) so the two
-    # janitors don't issue parallel CF API bursts at the same minute.
-    - cron: '45 * * * *'
-  workflow_dispatch:
-    inputs:
-      dry_run:
-        description: "Dry run only — list what would be deleted, no deletion"
-        required: false
-        type: boolean
-        default: true
-      max_delete_pct:
-        description: "Override safety gate (default 90, set higher only for major cleanup)"
-        required: false
-        default: "90"
-
-# Don't let two sweeps race the same account.
-concurrency:
-  group: sweep-cf-tunnels
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  sweep:
-    name: Sweep CF tunnels
-    runs-on: ubuntu-latest
-    # 30 min cap. Was 5 min on the theory that the only thing that
-    # could take >5min is a CF-API hang — but on 2026-05-02 a backlog
-    # of 672 stale tunnels accumulated (large staging E2E run + delayed
-    # sweep) and the serial `curl -X DELETE` loop (~0.7s/tunnel) needed
-    # ~7-8min to drain. The 5-min cap killed the run mid-sweep
-    # (cancelled at 424/672, see run 25248788312); a manual rerun
-    # finished the remainder fine.
-    #
-    # The fix is two-part: parallelize the delete loop (8-way xargs in
-    # the script — see scripts/ops/sweep-cf-tunnels.sh), AND raise the
-    # cap so a one-off backlog doesn't trip a hangs-detector that
-    # turned out to be a real-job-too-slow detector. With 8-way
-    # parallelism, 600+ tunnels drains in ~60s; 30 min is generous
-    # headroom for actual hangs to still surface (and is in line with
-    # the sweep-cf-orphans companion job).
-    timeout-minutes: 30
-    env:
-      CF_API_TOKEN: ${{ secrets.CF_API_TOKEN }}
-      CF_ACCOUNT_ID: ${{ secrets.CF_ACCOUNT_ID }}
-      CP_PROD_ADMIN_TOKEN: ${{ secrets.CP_PROD_ADMIN_TOKEN }}
-      CP_STAGING_ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_TOKEN }}
-      MAX_DELETE_PCT: ${{ github.event.inputs.max_delete_pct || '90' }}
-
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-
-      - name: Verify required secrets present
-        id: verify
-        # Schedule-vs-dispatch behaviour split mirrors sweep-cf-orphans
-        # (hardened 2026-04-28 after the silent-no-op incident: the
-        # janitor reported green while doing nothing because secrets
-        # were unset, masking a 152/200 zone-record leak). Same
-        # principle applies here:
-        #   - schedule → exit 1 on missing secrets (red CI surfaces it)
-        #   - workflow_dispatch → exit 0 with warning (operator-driven,
-        #     they already accepted the repo state)
-        run: |
-          missing=()
-          for var in CF_API_TOKEN CF_ACCOUNT_ID CP_PROD_ADMIN_TOKEN CP_STAGING_ADMIN_TOKEN; do
-            if [ -z "${!var:-}" ]; then
-              missing+=("$var")
-            fi
-          done
-          if [ ${#missing[@]} -gt 0 ]; then
-            if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-              echo "::warning::skipping sweep — secrets not configured: ${missing[*]}"
-              echo "::warning::set them at Settings → Secrets and Variables → Actions, then rerun."
-              echo "::warning::CF_API_TOKEN must include account:cloudflare_tunnel:edit scope (separate from the zone:dns:edit scope used by sweep-cf-orphans)."
-              echo "skip=true" >> "$GITHUB_OUTPUT"
-              exit 0
-            fi
-            echo "::error::sweep cannot run — required secrets missing: ${missing[*]}"
-            echo "::error::set them at Settings → Secrets and Variables → Actions, or disable this workflow."
-            echo "::error::CF_API_TOKEN must include account:cloudflare_tunnel:edit scope."
-            exit 1
-          fi
-          echo "All required secrets present ✓"
-          echo "skip=false" >> "$GITHUB_OUTPUT"
-
-      - name: Run sweep
-        if: steps.verify.outputs.skip != 'true'
-        # Schedule-vs-dispatch dry-run asymmetry mirrors sweep-cf-orphans:
-        #   - Scheduled: input empty → "false" → --execute (the whole
-        #     point of an hourly janitor).
-        #   - Manual workflow_dispatch: input default true → dry-run;
-        #     operator must flip it to actually delete.
-        run: |
-          set -euo pipefail
-          if [ "${{ github.event.inputs.dry_run || 'false' }}" = "true" ]; then
-            echo "Running in dry-run mode — no deletions"
-            bash scripts/ops/sweep-cf-tunnels.sh
-          else
-            echo "Running with --execute — will delete identified orphans"
-            bash scripts/ops/sweep-cf-tunnels.sh --execute
-          fi
@@ -1,239 +0,0 @@
-name: Sweep stale e2e-* orgs (staging)
-
-# Janitor for staging tenants left behind when E2E cleanup didn't run:
-# CI cancellations, runner crashes, transient AWS errors mid-cascade,
-# bash trap missed (signal 9), etc. Without this loop, every failed
-# teardown leaks an EC2 + DNS + DB row until manual ops cleanup —
-# 2026-04-23 staging hit the 64 vCPU AWS quota from ~27 such orphans.
-#
-# Why not rely on per-test-run teardown:
-#   - Per-run teardown is best-effort by definition. Any process death
-#     after the test starts but before the trap fires leaves debris.
-#   - GH Actions cancellation kills the runner without grace period.
-#     The workflow's `if: always()` step usually catches this, but it
-#     too can fail (CP transient 5xx, runner network issue at the
-#     wrong moment).
-#   - Even when teardown runs, the CP cascade is best-effort in places
-#     (cascadeTerminateWorkspaces logs+continues; DNS deletion same).
-#   - This sweep is the catch-all that converges staging back to clean
-#     regardless of which specific path leaked.
-#
-# The PROPER fix is making CP cleanup transactional + verify-after-
-# terminate (filed separately as cleanup-correctness work). This
-# workflow is the safety net that catches everything else AND any
-# future leak source we haven't yet identified.
-
-on:
-  schedule:
-    # Every 15 min. E2E orgs are short-lived (~8-25 min wall clock from
-    # create to teardown — canary is ~8 min, full SaaS ~25 min). The
-    # previous hourly + 120-min stale threshold meant a leaked tenant
-    # could keep an EC2 alive for up to 2 hours, eating ~2 vCPU per
-    # leak. Tightening the cadence + threshold reduces the worst-case
-    # leak window from 120 min to ~45 min (15-min sweep cadence + 30-min
-    # threshold) without risk of catching in-progress runs (the longest
-    # e2e run is the 25-min canary, well under the 30-min threshold).
-    # See molecule-controlplane#420 for the leak-class accounting that
-    # motivated this tightening.
-    - cron: '*/15 * * * *'
-  workflow_dispatch:
-    inputs:
-      max_age_minutes:
-        description: "Delete e2e-* orgs older than N minutes (default 30)"
-        required: false
-        default: "30"
-      dry_run:
-        description: "Dry run only — list what would be deleted"
-        required: false
-        type: boolean
-        default: false
-
-# Don't let two sweeps fight. Cron + workflow_dispatch could overlap
-# on a manual trigger; queue rather than parallel-delete.
-concurrency:
-  group: sweep-stale-e2e-orgs
-  cancel-in-progress: false
-
-permissions:
-  contents: read
-
-jobs:
-  sweep:
-    name: Sweep e2e orgs
-    runs-on: ubuntu-latest
-    timeout-minutes: 15
-    env:
-      MOLECULE_CP_URL: https://staging-api.moleculesai.app
-      ADMIN_TOKEN: ${{ secrets.MOLECULE_STAGING_ADMIN_TOKEN }}
-      MAX_AGE_MINUTES: ${{ github.event.inputs.max_age_minutes || '30' }}
-      DRY_RUN: ${{ github.event.inputs.dry_run || 'false' }}
-      # Refuse to delete more than this many orgs in one tick. If the
-      # CP DB is briefly empty (or the admin endpoint goes weird and
-      # returns no created_at), every e2e- org would look stale.
-      # Bailing protects against runaway nukes.
-      SAFETY_CAP: 50
-
-    steps:
-      - name: Verify admin token present
-        run: |
-          if [ -z "$ADMIN_TOKEN" ]; then
-            echo "::error::MOLECULE_STAGING_ADMIN_TOKEN not set"
-            exit 2
-          fi
-          echo "Admin token present ✓"
-
-      - name: Identify stale e2e orgs
-        id: identify
-        run: |
-          set -euo pipefail
-          # Fetch into a file so the python step reads it via stdin —
-          # cleaner than embedding $(curl ...) into a heredoc.
-          curl -sS --fail-with-body --max-time 30 \
-            "$MOLECULE_CP_URL/cp/admin/orgs?limit=500" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" \
-            > orgs.json
-
-          # Filter:
-          #   1. slug starts with one of the ephemeral test prefixes:
-          #        - 'e2e-'    — covers e2e-canary-, e2e-canvas-*, etc.
-          #        - 'rt-e2e-' — runtime-test harness fixtures (RFC #2251);
-          #                      missing this prefix left two such tenants
-          #                      orphaned 8h on staging (2026-05-03), then
-          #                      hard-failed redeploy-tenants-on-staging
-          #                      and broke the staging→main auto-promote
-          #                      chain. Kept in sync with the EPHEMERAL_PREFIX_RE
-          #                      regex in redeploy-tenants-on-staging.yml.
-          #   2. created_at is older than MAX_AGE_MINUTES ago
-          # Output one slug per line to a file the next step reads.
-          python3 > stale_slugs.txt <<'PY'
-          import json, os
-          from datetime import datetime, timezone, timedelta
-          # SSOT for this list lives in the controlplane Go code:
-          # molecule-controlplane/internal/slugs/ephemeral.go
-          # (var EphemeralPrefixes). The redeploy-fleet auto-rollout
-          # also reads from there to SKIP these slugs — without that
-          # filter, fleet redeploy SSM-failed in-flight E2E tenants
-          # whose containers were still booting, breaking the test
-          # that just spun them up (molecule-controlplane#493).
-          # Update both files together.
-          EPHEMERAL_PREFIXES = ("e2e-", "rt-e2e-")
-          with open("orgs.json") as f:
-              data = json.load(f)
-          max_age = int(os.environ["MAX_AGE_MINUTES"])
-          cutoff = datetime.now(timezone.utc) - timedelta(minutes=max_age)
-          for o in data.get("orgs", []):
-              slug = o.get("slug", "")
-              if not slug.startswith(EPHEMERAL_PREFIXES):
-                  continue
-              created = o.get("created_at")
-              if not created:
-                  # Defensively skip rows without created_at — better
-                  # to leave one orphan than nuke a brand-new row
-                  # whose timestamp didn't render.
-                  continue
-              # Python 3.11+ handles RFC3339 with Z directly via
-              # fromisoformat; older runners need the trailing Z swap.
-              created_dt = datetime.fromisoformat(created.replace("Z", "+00:00"))
-              if created_dt < cutoff:
-                  print(slug)
-          PY
-
-          count=$(wc -l < stale_slugs.txt | tr -d ' ')
-          echo "Found $count stale e2e org(s) older than ${MAX_AGE_MINUTES}m"
-          if [ "$count" -gt 0 ]; then
-            echo "First 20:"
-            head -20 stale_slugs.txt | sed 's/^/  /'
-          fi
-          echo "count=$count" >> "$GITHUB_OUTPUT"
-
-      - name: Safety gate
-        if: steps.identify.outputs.count != '0'
-        run: |
-          count="${{ steps.identify.outputs.count }}"
-          if [ "$count" -gt "$SAFETY_CAP" ]; then
-            echo "::error::Refusing to delete $count orgs in one sweep (cap=$SAFETY_CAP). Investigate manually — this usually means the CP admin API returned no created_at or returned a degraded result. Re-run with workflow_dispatch + max_age_minutes if intentional."
-            exit 1
-          fi
-          echo "Within safety cap ($count ≤ $SAFETY_CAP) ✓"
-
-      - name: Delete stale orgs
-        if: steps.identify.outputs.count != '0' && env.DRY_RUN != 'true'
-        run: |
-          set -uo pipefail
-          deleted=0
-          failed=0
-          while IFS= read -r slug; do
-            [ -z "$slug" ] && continue
-            # The DELETE handler requires {"confirm": "<slug>"} matching
-            # the URL slug — fat-finger guard. Idempotent: re-issuing
-            # picks up via org_purges.last_step.
-            # Tempfile-routed -w + set +e/-e prevents curl-exit-code
-            # pollution of the captured status (lint-curl-status-capture.yml).
-            set +e
-            curl -sS -o /tmp/del_resp -w "%{http_code}" \
-              --max-time 60 \
-              -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
-              -H "Authorization: Bearer $ADMIN_TOKEN" \
-              -H "Content-Type: application/json" \
-              -d "{\"confirm\":\"$slug\"}" >/tmp/del_code
-            set -e
-            # Stderr from curl (-sS shows dial errors etc.) goes to runner log.
-            http_code=$(cat /tmp/del_code 2>/dev/null || echo "000")
-            if [ "$http_code" = "200" ] || [ "$http_code" = "204" ]; then
-              deleted=$((deleted+1))
-              echo "  deleted: $slug"
-            else
-              failed=$((failed+1))
-              echo "  FAILED ($http_code): $slug — $(cat /tmp/del_resp 2>/dev/null | head -c 200)"
-            fi
-          done < stale_slugs.txt
-          echo ""
-          echo "Sweep summary: deleted=$deleted failed=$failed"
-          # Don't fail the workflow on per-org delete errors — the
-          # sweeper is best-effort. Next hourly tick re-attempts. We
-          # only fail loud at the safety-cap gate above.
-
-      - name: Sweep orphan tunnels
-        # Stale-org cleanup deletes the org (which cascades to tunnel
-        # delete inside the CP). But when that cascade fails partway —
-        # CP transient 5xx after the org row is deleted but before the
-        # CF tunnel delete completes — the tunnel persists with no
-        # matching org row. The reconciler in internal/sweep flags this
-        # as `cf_tunnel kind=orphan`, but nothing automatically reaps it.
-        #
-        # `/cp/admin/orphan-tunnels/cleanup` is the operator-triggered
-        # reaper. Calling it here at the end of every sweep tick
-        # converges the staging CF account to clean even when CP
-        # cascades half-fail.
-        #
-        # PR #492 made the underlying DeleteTunnel actually check
-        # status — pre-fix it silent-succeeded on CF code 1022
-        # ("active connections"), so this step would have been a no-op
-        # against stuck connectors. Post-fix the cleanup invokes
-        # CleanupTunnelConnections + retry, which actually clears the
-        # 1022 case. (#2987)
-        #
-        # Best-effort. Failure here doesn't fail the workflow — next
-        # tick re-attempts. Errors flow to step output for ops review.
-        if: env.DRY_RUN != 'true'
-        run: |
-          set +e
-          curl -sS -o /tmp/cleanup_resp -w "%{http_code}" \
-            --max-time 60 \
-            -X POST "$MOLECULE_CP_URL/cp/admin/orphan-tunnels/cleanup" \
-            -H "Authorization: Bearer $ADMIN_TOKEN" >/tmp/cleanup_code
-          set -e
-          http_code=$(cat /tmp/cleanup_code 2>/dev/null || echo "000")
-          body=$(cat /tmp/cleanup_resp 2>/dev/null | head -c 500)
-          if [ "$http_code" = "200" ]; then
-            count=$(echo "$body" | python3 -c "import sys,json; d=json.loads(sys.stdin.read() or '{}'); print(d.get('deleted_count', 0))" 2>/dev/null || echo "0")
-            failed_n=$(echo "$body" | python3 -c "import sys,json; d=json.loads(sys.stdin.read() or '{}'); print(len(d.get('failed') or {}))" 2>/dev/null || echo "0")
-            echo "Orphan-tunnel sweep: deleted=$count failed=$failed_n"
-          else
-            echo "::warning::orphan-tunnels cleanup returned HTTP $http_code — body: $body"
-          fi
-
-      - name: Dry-run summary
-        if: env.DRY_RUN == 'true'
-        run: |
-          echo "DRY RUN — would have deleted ${{ steps.identify.outputs.count }} org(s) AND triggered orphan-tunnels cleanup. Re-run with dry_run=false to actually delete."
@@ -1,52 +0,0 @@
-name: Ops Scripts Tests
-
-# Runs the unittest suite for scripts/ on every PR + push that touches
-# anything under scripts/. Kept separate from the main CI so a script-only
-# change doesn't trigger the heavier Go/Canvas/Python pipelines.
-#
-# Discovery layout: tests sit alongside the code they test (see
-# scripts/ops/test_sweep_cf_decide.py for the pattern; scripts/
-# test_build_runtime_package.py for the rewriter coverage). The job
-# below runs `unittest discover` TWICE — once from `scripts/`, once
-# from `scripts/ops/` — because neither dir has an `__init__.py`, so
-# a single discover from `scripts/` doesn't recurse into the ops
-# subdir. Two passes is simpler than retrofitting namespace packages.
-
-on:
-  push:
-    branches: [main, staging]
-    paths:
-      - 'scripts/**'
-      - '.github/workflows/test-ops-scripts.yml'
-  pull_request:
-    branches: [main, staging]
-    paths:
-      - 'scripts/**'
-      - '.github/workflows/test-ops-scripts.yml'
-  merge_group:
-    types: [checks_requested]
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  test:
-    name: Ops scripts (unittest)
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
-        with:
-          python-version: '3.11'
-      - name: Run scripts/ unittests (build_runtime_package, …)
-        # Top-level scripts/ tests live alongside their target file
-        # (e.g. scripts/test_build_runtime_package.py exercises
-        # scripts/build_runtime_package.py). discover from scripts/
-        # picks up only top-level test_*.py because scripts/ops/ has
-        # no __init__.py — that's intentional, so we run two passes.
-        working-directory: scripts
-        run: python -m unittest discover -t . -p 'test_*.py' -v
-      - name: Run scripts/ops/ unittests (sweep_cf_decide, …)
-        working-directory: scripts/ops
-        run: python -m unittest discover -p 'test_*.py' -v
@@ -57,24 +57,24 @@ See `CLAUDE.md` for a full list of environment variables and their purposes.

 This repo is scoped to **code** (canvas, workspace, workspace-server, related
 infra). Public content (blog posts, marketing copy, OG images, SEO briefs,
-DevRel demos) lives in [`Molecule-AI/docs`](https://git.moleculesai.app/molecule-ai/docs).
+DevRel demos) lives in [`molecule-ai/docs`](https://git.moleculesai.app/molecule-ai/docs).
 The `Block forbidden paths` CI gate fails any PR that writes to `marketing/`
-or other removed paths — open against `Molecule-AI/docs` instead.
+or other removed paths — open against `molecule-ai/docs` instead.

 | Content type | Target |
 |---|---|
-| Blog posts | `Molecule-AI/docs` → `content/blog/<YYYY-MM-DD-slug>/` |
-| Doc pages | `Molecule-AI/docs` → `content/docs/` |
-| Marketing copy / PMM positioning | `Molecule-AI/docs` → `marketing/` |
-| OG images, visual assets | `Molecule-AI/docs` → `app/` or `marketing/` |
-| SEO briefs | `Molecule-AI/docs` → `marketing/` |
-| DevRel demos (runnable code) | Standalone repo under `Molecule-AI/`, OR embedded in `Molecule-AI/docs` |
-| Launch checklists, internal tracking | GitHub Issues — **not** committed files |
+| Blog posts | `molecule-ai/docs` → `content/blog/<YYYY-MM-DD-slug>/` |
+| Doc pages | `molecule-ai/docs` → `content/docs/` |
+| Marketing copy / PMM positioning | `molecule-ai/docs` → `marketing/` |
+| OG images, visual assets | `molecule-ai/docs` → `app/` or `marketing/` |
+| SEO briefs | `molecule-ai/docs` → `marketing/` |
+| DevRel demos (runnable code) | Standalone repo under `molecule-ai/`, OR embedded in `molecule-ai/docs` |
+| Launch checklists, internal tracking | Gitea Issues — **not** committed files |
 | Engineering docs (`docs/adr/`, `docs/architecture/`, `docs/incidents/`) | This repo (internal, not published) |
 | Live product pages (e.g. `canvas/src/app/pricing/page.tsx`) | This repo (these are app code, not marketing copy) |

 If a PR fails the `Block forbidden paths` check, the contents belong in
-`Molecule-AI/docs`. No CI drag, no Canvas E2E, content lands in minutes.
+`molecule-ai/docs`. No CI drag, no Canvas E2E, content lands in minutes.

 ## Development Workflow

@@ -106,7 +106,7 @@ causing a render loop when any node position changed.

 #### Auto-merge & the "extra commit" trap

-**Two system guards protect against pushing commits after auto-merge has been enabled.** Don't try to work around them — they exist because we shipped a half-merged PR on 2026-04-27 (`#2174` merged with only its first commit; the second was orphaned on a branch GitHub had already deleted).
+**Two system guards protect against pushing commits after auto-merge has been enabled.** Don't try to work around them — they exist because we shipped a half-merged PR on 2026-04-27 (`#2174` merged with only its first commit; the second was orphaned on a branch the host had already deleted).

 1. **Repo-wide:** "Automatically delete head branches" is on. Once a PR merges, the branch is deleted server-side. Any subsequent `git push` to that branch fails with `remote rejected — no such branch`.

@@ -145,7 +145,7 @@ Fix violations before committing — the hook will reject the commit.

 ### CI Pipeline

-CI runs on GitHub Actions with a self-hosted runner. External contributors:
+CI runs on Gitea Actions with self-hosted runners. External contributors:
 PRs from forks will not trigger CI automatically. A maintainer will review
 and run CI manually.

@@ -190,9 +190,9 @@ Runs the full regression suite against a fixture HTTP server. No network access
 Code in this repo lands in molecule-core. Some related runtime artifacts
 live in their own repos:

- [`Molecule-AI/molecule-ai-workspace-runtime`](https://git.moleculesai.app/molecule-ai/molecule-ai-workspace-runtime) — Python adapter SDK (`molecule_runtime`) that runs inside containerized Molecule workspaces. Bridges Claude Code SDK / hermes / langgraph / etc. → A2A queue.
- [`Molecule-AI/molecule-sdk-python`](https://git.moleculesai.app/molecule-ai/molecule-sdk-python) — `A2AServer` + `RemoteAgentClient` for external agents that register over the public `/registry/register` flow.
- [`Molecule-AI/molecule-mcp-claude-channel`](https://git.moleculesai.app/molecule-ai/molecule-mcp-claude-channel) — Claude Code channel plugin. Bridges A2A traffic into a running Claude Code session via MCP `notifications/claude/channel`. Polling-based (no tunnel required); install with `claude --channels plugin:molecule@Molecule-AI/molecule-mcp-claude-channel`.
+- [`molecule-ai/molecule-ai-workspace-runtime`](https://git.moleculesai.app/molecule-ai/molecule-ai-workspace-runtime) — Python adapter SDK (`molecule_runtime`) that runs inside containerized Molecule workspaces. Bridges Claude Code SDK / hermes / langgraph / etc. → A2A queue.
+- [`molecule-ai/molecule-sdk-python`](https://git.moleculesai.app/molecule-ai/molecule-sdk-python) — `A2AServer` + `RemoteAgentClient` for external agents that register over the public `/registry/register` flow.
+- [`molecule-ai/molecule-mcp-claude-channel`](https://git.moleculesai.app/molecule-ai/molecule-mcp-claude-channel) — Claude Code channel plugin. Bridges A2A traffic into a running Claude Code session via MCP `notifications/claude/channel`. Polling-based (no tunnel required); install inside Claude Code via `/plugin marketplace add https://git.moleculesai.app/molecule-ai/molecule-mcp-claude-channel.git` → `/plugin install molecule@molecule-channel`, then launch with `claude --dangerously-load-development-channels=plugin:molecule@molecule-channel`.

 When extending the **A2A surface** in molecule-core (`workspace-server/internal/handlers/a2a_proxy.go` etc.), consider whether the change has a downstream impact on the runtime SDK or the channel plugin — they're versioned independently but share the wire shape.

@@ -206,7 +206,7 @@ See `CLAUDE.md` for detailed architecture documentation, including:

 ## Reporting Issues

-Use GitHub Issues with a clear title and reproduction steps. Include:
+Use Gitea Issues with a clear title and reproduction steps. Include:
 - What you expected
 - What actually happened
 - Platform/OS version
@@ -214,8 +214,9 @@ Use GitHub Issues with a clear title and reproduction steps. Include:

 ## Security

-If you discover a security vulnerability, please report it privately via
-GitHub Security Advisories rather than opening a public issue.
+If you discover a security vulnerability, please report it privately by
+opening an issue against `molecule-ai/internal` (a private repo only
+maintainers can see) rather than filing a public issue here.

 ## License

@@ -4,10 +4,10 @@
 # use this Makefile; CI calls docker compose / go test directly so the
 # Makefile can evolve without breaking the build.

-.PHONY: help dev up down logs build test
+.PHONY: help dev up down logs build test e2e-peer-visibility

 help: ## Show this help.
-	@grep -E '^[a-zA-Z_-]+:.*?## ' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-12s\033[0m %s\n", $$1, $$2}'
+	@grep -E '^[a-zA-Z0-9_-]+:.*?## ' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-22s\033[0m %s\n", $$1, $$2}'

 dev: ## Start the full stack with air hot-reload for the platform service.
 	docker compose -f docker-compose.yml -f docker-compose.dev.yml up
@@ -26,3 +26,13 @@ build: ## Force a fresh build of the platform image (no cache).

 test: ## Run Go unit tests in workspace-server/.
 	cd workspace-server && go test -race ./...
+
+# ─── Local prod-mimic E2E gates ────────────────────────────────────────
+# Run the LITERAL peer-visibility MCP list_peers gate against the
+# already-running local stack (`make up` or `make dev`). Same byte-
+# identical assertion as the staging gate — only provisioning differs.
+# Skips any runtime whose provider key is absent (partially-keyed env
+# is fine). See tests/e2e/test_peer_visibility_mcp_local.sh for the
+# env contract (CLAUDE_CODE_OAUTH_TOKEN / E2E_MINIMAX_API_KEY / etc).
+e2e-peer-visibility: ## Run the LOCAL peer-visibility MCP gate vs the running stack (needs `make up` first).
+	bash tests/e2e/test_peer_visibility_mcp_local.sh
@@ -238,7 +238,7 @@ The result is not just “an agent that learns.” It is **an organization that
 - subscribe to one or more workspaces; peer messages surface as conversation turns; replies route back through Molecule's A2A
 - no tunnel, no public endpoint — the plugin self-registers each watched workspace as `delivery_mode=poll` and long-polls `/activity?since_id=…`
 - multi-tenant friendly: one plugin install can watch workspaces across multiple Molecule tenants (`MOLECULE_PLATFORM_URLS` per-workspace)
- install via the standard marketplace flow: `/plugin marketplace add Molecule-AI/molecule-mcp-claude-channel` → `/plugin install molecule-channel@molecule-mcp-claude-channel`
+- install via the standard marketplace flow: `/plugin marketplace add https://git.moleculesai.app/molecule-ai/molecule-mcp-claude-channel.git` → `/plugin install molecule@molecule-channel`, then launch with `claude --dangerously-load-development-channels=plugin:molecule@molecule-channel`

 ## Built For Teams That Need More Than A Demo

@@ -237,7 +237,7 @@ Molecule AI 并不是要替代下面这些 framework，而是把它们纳入更
 - 订阅一个或多个 workspace；peer 的消息会以 user-turn 出现，回复会经 Molecule A2A 路由出去
 - 无需公网隧道、无需公开端点 —— 插件启动时自动把每个 watched workspace 注册成 `delivery_mode=poll`，长轮询 `/activity?since_id=…`
 - 多租户友好：单次安装即可同时 watch 跨多个 Molecule 租户的 workspace（`MOLECULE_PLATFORM_URLS` 按 workspace 配置）
- 通过标准 marketplace 流程安装：`/plugin marketplace add Molecule-AI/molecule-mcp-claude-channel` → `/plugin install molecule-channel@molecule-mcp-claude-channel`
+- 通过标准 marketplace 流程安装：`/plugin marketplace add https://git.moleculesai.app/molecule-ai/molecule-mcp-claude-channel.git` → `/plugin install molecule@molecule-channel`，然后用 `claude --dangerously-load-development-channels=plugin:molecule@molecule-channel` 启动

 ## 适合什么团队

@@ -1 +1,2 @@
-trigger
+trigger
+retrigger 2026-05-20T04:09Z after op-config#110 (HOME=/home/runner) deploy to fleet — internal#603
@@ -0,0 +1,113 @@
+import { describe, it, expect, vi } from "vitest";
+
+// Marketing-launch SEO (mc#1486). These tests pin the public crawler
+// contract: anything that flips public marketing routes to disallow,
+// drops the sitemap from robots.txt, or removes the OG image
+// reference from root metadata should fail loudly here.
+
+// next/font and the rest of the layout's runtime tree are not
+// vitest-compatible (next/font expects the Next.js compiler swc
+// transform). We import layout.tsx only for its exported `metadata`
+// constant — mock the font module to a constructor-returning stub.
+vi.mock("next/font/google", () => ({
+  Inter: () => ({ variable: "--font-inter" }),
+  JetBrains_Mono: () => ({ variable: "--font-jetbrains" }),
+}));
+
+import robots from "../robots";
+import sitemap from "../sitemap";
+import { metadata } from "../layout";
+
+describe("robots.ts", () => {
+  it("allows public marketing routes and blocks authed/app routes", () => {
+    const r = robots();
+    expect(r.rules).toBeDefined();
+    const rule = Array.isArray(r.rules) ? r.rules[0] : r.rules!;
+    expect(rule.userAgent).toBe("*");
+    const allow = Array.isArray(rule.allow) ? rule.allow : [rule.allow];
+    expect(allow).toEqual(expect.arrayContaining(["/", "/pricing", "/blog"]));
+    const disallow = Array.isArray(rule.disallow)
+      ? rule.disallow
+      : [rule.disallow];
+    expect(disallow).toEqual(
+      expect.arrayContaining(["/api/", "/orgs", "/cp/"]),
+    );
+  });
+
+  it("declares the sitemap URL", () => {
+    const r = robots();
+    expect(r.sitemap).toMatch(/\/sitemap\.xml$/);
+  });
+
+  it("declares a canonical host", () => {
+    const r = robots();
+    expect(r.host).toMatch(/^https:\/\//);
+  });
+});
+
+describe("sitemap.ts", () => {
+  it("includes apex, pricing, and the live blog post", () => {
+    const entries = sitemap();
+    const urls = entries.map((e) => e.url);
+    expect(urls.some((u) => u.endsWith("/"))).toBe(true);
+    expect(urls.some((u) => u.endsWith("/pricing"))).toBe(true);
+    expect(
+      urls.some((u) => u.includes("/blog/2026-04-20-chrome-devtools-mcp")),
+    ).toBe(true);
+  });
+
+  it("does NOT include authed/app routes", () => {
+    const entries = sitemap();
+    const urls = entries.map((e) => e.url);
+    expect(urls.some((u) => u.includes("/orgs"))).toBe(false);
+    expect(urls.some((u) => u.includes("/api/"))).toBe(false);
+  });
+
+  it("sets a non-zero priority and a valid changeFrequency on every entry", () => {
+    const valid = new Set([
+      "always",
+      "hourly",
+      "daily",
+      "weekly",
+      "monthly",
+      "yearly",
+      "never",
+    ]);
+    for (const e of sitemap()) {
+      expect(e.priority).toBeGreaterThan(0);
+      expect(valid.has(String(e.changeFrequency))).toBe(true);
+    }
+  });
+});
+
+describe("root layout metadata", () => {
+  it("sets a templated title + non-empty description", () => {
+    const t = metadata.title as { default: string; template: string };
+    expect(t.default).toMatch(/Molecule AI/);
+    expect(t.template).toMatch(/%s/);
+    expect((metadata.description ?? "").length).toBeGreaterThan(50);
+  });
+
+  it("declares OG + Twitter text fields (image comes from opengraph-image.tsx)", () => {
+    const og = metadata.openGraph;
+    expect(og).toBeDefined();
+    expect((og as { title: string }).title).toMatch(/Molecule AI/);
+    expect((og as { description: string }).description.length).toBeGreaterThan(
+      50,
+    );
+    const tw = metadata.twitter;
+    expect(tw).toBeDefined();
+    // Next.js typings narrow twitter.card to a union — assert via cast.
+    expect((tw as { card: string }).card).toBe("summary_large_image");
+  });
+
+  it("sets a canonical alternate", () => {
+    expect(metadata.alternates?.canonical).toBe("/");
+  });
+
+  it("enables indexing at the metadata level (robots.ts owns per-route)", () => {
+    const r = metadata.robots as { index: boolean; follow: boolean };
+    expect(r.index).toBe(true);
+    expect(r.follow).toBe(true);
+  });
+});
@@ -27,9 +27,78 @@ import {
  themeBootScript,
 } from "@/lib/theme-cookie";

+// Marketing-launch SEO (mc#1486). Canonical apex is app.moleculesai.app —
+// tenant subdomains (<slug>.moleculesai.app) reuse the same Next.js build
+// but are gated behind auth (AuthGate redirects anonymous → /cp/auth/login)
+// and are de-indexed in robots.ts. The metadata here applies to the
+// public marketing surface served from the apex host.
+//
+// Override per-route by exporting a page-level `metadata`/`generateMetadata`
+// — Next.js merges page metadata over layout metadata using
+// `title.template` for "<page> | Molecule AI" composition.
+const SITE_URL =
+  process.env.NEXT_PUBLIC_SITE_URL ?? "https://app.moleculesai.app";
+
 export const metadata: Metadata = {
-  title: "Molecule AI",
-  description: "AI Org Chart Canvas",
+  metadataBase: new URL(SITE_URL),
+  title: {
+    default: "Molecule AI — the AI org chart canvas",
+    template: "%s | Molecule AI",
+  },
+  description:
+    "Molecule AI is an org-chart canvas for AI agent teams. Wire Claude Code, Codex, Hermes, and OpenClaw agents into a governed multi-agent workspace with credit metering, audit, and one-click runtime provisioning.",
+  applicationName: "Molecule AI",
+  keywords: [
+    "AI agents",
+    "multi-agent",
+    "agent orchestration",
+    "AI org chart",
+    "Claude Code",
+    "Codex",
+    "MCP",
+    "agent governance",
+    "A2A",
+    "agent runtime",
+  ],
+  authors: [{ name: "Molecule AI" }],
+  creator: "Molecule AI",
+  publisher: "Molecule AI",
+  alternates: { canonical: "/" },
+  // OG + Twitter images come from the file-convention sibling
+  // `opengraph-image.tsx` — Next.js auto-attaches them to og:image
+  // and twitter:image when present at the segment root. We keep the
+  // text fields here so they win over per-page metadata when a page
+  // doesn't override them. `images: []` as the structural fallback
+  // for hosts that won't follow the file convention; the real URL
+  // is injected by Next.js at build time from opengraph-image.tsx.
+  openGraph: {
+    type: "website",
+    siteName: "Molecule AI",
+    url: SITE_URL,
+    title: "Molecule AI — the AI org chart canvas",
+    description:
+      "Wire Claude Code, Codex, Hermes, and OpenClaw agents into a governed multi-agent workspace. Credit metering, audit, and one-click runtime provisioning.",
+    locale: "en_US",
+  },
+  twitter: {
+    card: "summary_large_image",
+    title: "Molecule AI — the AI org chart canvas",
+    description:
+      "Wire Claude Code, Codex, Hermes, and OpenClaw agents into a governed multi-agent workspace.",
+  },
+  icons: {
+    icon: "/molecule-icon.png",
+    apple: "/molecule-icon.png",
+  },
+  // robots.ts owns the per-route allow/disallow contract; this is the
+  // header-level fallback for routes the crawler reaches before
+  // robots.txt resolves. Default = index public marketing routes;
+  // app/auth/api/orgs are noindex'd by robots.ts.
+  robots: {
+    index: true,
+    follow: true,
+    googleBot: { index: true, follow: true, "max-image-preview": "large" },
+  },
 };

 export default async function RootLayout({
@@ -94,6 +163,75 @@ export default async function RootLayout({
          nonce={nonce}
          dangerouslySetInnerHTML={{ __html: themeBootScript }}
        />
+        {/*
+         * JSON-LD structured data (mc#1486). Two graph nodes:
+         *
+         *   - Organization: surfaces the brand to Google Knowledge
+         *     Graph + Bing entity index. URL+logo+sameAs are the
+         *     minimum recommended set for new brands without a
+         *     Wikipedia page.
+         *
+         *   - WebSite: enables the sitelinks search box and tells
+         *     crawlers the canonical site URL when the same content
+         *     is reachable via multiple subdomains (apex + tenant).
+         *
+         * Type-application/ld+json runs synchronously without
+         * executing JS, so 'strict-dynamic' isn't required — we still
+         * carry the nonce because production CSP's default-src 'self'
+         * applies to any <script> element. The "type" attribute is
+         * what keeps the browser from running the body as JS, but
+         * CSP nonces are gated on the element not the type, so we
+         * include the nonce too.
+         */}
+        <script
+          type="application/ld+json"
+          nonce={nonce}
+          dangerouslySetInnerHTML={{
+            __html: JSON.stringify({
+              "@context": "https://schema.org",
+              "@graph": [
+                {
+                  "@type": "Organization",
+                  "@id": `${SITE_URL}#organization`,
+                  name: "Molecule AI",
+                  url: SITE_URL,
+                  logo: `${SITE_URL}/molecule-icon.png`,
+                  sameAs: [
+                    "https://github.com/molecule-ai",
+                    "https://x.com/moleculeai",
+                  ],
+                },
+                {
+                  "@type": "WebSite",
+                  "@id": `${SITE_URL}#website`,
+                  url: SITE_URL,
+                  name: "Molecule AI",
+                  publisher: { "@id": `${SITE_URL}#organization` },
+                  inLanguage: "en-US",
+                },
+                {
+                  "@type": "SoftwareApplication",
+                  "@id": `${SITE_URL}#software`,
+                  name: "Molecule AI",
+                  applicationCategory: "DeveloperApplication",
+                  operatingSystem: "Web",
+                  description:
+                    "Org-chart canvas for AI agent teams with credit metering, audit, and one-click runtime provisioning.",
+                  url: SITE_URL,
+                  offers: {
+                    "@type": "AggregateOffer",
+                    priceCurrency: "USD",
+                    lowPrice: "0",
+                    highPrice: "99",
+                    offerCount: "3",
+                    url: `${SITE_URL}/pricing`,
+                  },
+                  publisher: { "@id": `${SITE_URL}#organization` },
+                },
+              ],
+            }),
+          }}
+        />
      </head>
      <body className={`bg-surface text-ink ${interFont.variable} ${monoFont.variable}`}>
        <ThemeProvider initialTheme={theme}>
@@ -0,0 +1,82 @@
+import { ImageResponse } from "next/og";
+
+// Marketing-launch SEO (mc#1486). Next.js App-Router file-system OG
+// convention: served as `/opengraph-image` and auto-attached as
+// `og:image` + `twitter:image`. Dynamic (not a static PNG in /public)
+// so we can iterate the brand mark + tagline pre-launch without
+// churning a binary blob in git history.
+export const runtime = "edge";
+
+export const alt = "Molecule AI — the AI org chart canvas";
+export const size = { width: 1200, height: 630 };
+export const contentType = "image/png";
+
+export default function OG() {
+  return new ImageResponse(
+    (
+      <div
+        style={{
+          width: "100%",
+          height: "100%",
+          display: "flex",
+          flexDirection: "column",
+          alignItems: "flex-start",
+          justifyContent: "center",
+          padding: "80px",
+          background:
+            "linear-gradient(135deg, #0a0a0a 0%, #1a1a2e 60%, #16213e 100%)",
+          color: "#ffffff",
+          fontFamily: "system-ui, -apple-system, sans-serif",
+        }}
+      >
+        <div
+          style={{
+            fontSize: 28,
+            color: "#a3a3c2",
+            letterSpacing: "0.18em",
+            textTransform: "uppercase",
+            marginBottom: 24,
+          }}
+        >
+          Molecule AI
+        </div>
+        <div
+          style={{
+            fontSize: 76,
+            fontWeight: 700,
+            lineHeight: 1.05,
+            letterSpacing: "-0.02em",
+            maxWidth: 980,
+          }}
+        >
+          The AI org chart canvas
+        </div>
+        <div
+          style={{
+            fontSize: 32,
+            color: "#c8c8d8",
+            marginTop: 32,
+            lineHeight: 1.3,
+            maxWidth: 980,
+          }}
+        >
+          Wire Claude Code, Codex, Hermes, and OpenClaw agents into a governed
+          multi-agent workspace.
+        </div>
+        <div
+          style={{
+            position: "absolute",
+            right: 80,
+            bottom: 80,
+            fontSize: 22,
+            color: "#7a7a96",
+            display: "flex",
+          }}
+        >
+          moleculesai.app
+        </div>
+      </div>
+    ),
+    { ...size },
+  );
+}
@@ -103,7 +103,7 @@ export default function Home() {
                setHydrationError(null);
                window.location.reload();
              }}
-              className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm"
+              className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
            >
              Retry
            </button>
@@ -115,7 +115,9 @@ export default function Home() {

  return (
    <>
-      <Canvas />
+      <main aria-label="Agent canvas">
+        <Canvas />
+      </main>
      <Legend />
      <CommunicationOverlay />
      {hydrationError && (
@@ -134,7 +136,7 @@ export default function Home() {
              setHydrationError(null);
              window.location.reload();
            }}
-            className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm"
+            className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
          >
            Retry
          </button>
@@ -176,7 +178,7 @@ brew services start redis`}</pre>
      </p>
      <button
        onClick={() => window.location.reload()}
-        className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm mt-2"
+        className="px-4 py-2 bg-accent-strong hover:bg-accent text-white rounded-md text-sm mt-2 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
      >
        Reload
      </button>
@@ -0,0 +1,45 @@
+import type { MetadataRoute } from "next";
+
+// Marketing-launch SEO (mc#1486). Next.js App-Router robots convention:
+// this file is served as `/robots.txt` at build time and is the single
+// source of truth for crawler allow/disallow.
+//
+// Contract:
+//   - Public marketing routes (/, /pricing, /blog/*) are crawlable.
+//   - Authed/app routes (/orgs, /api/*) are noindex'd. They render
+//     useful content only after a session round-trip, so a crawler hit
+//     just wastes our crawl budget and exposes endpoint shapes.
+//   - Tenant subdomains (<slug>.moleculesai.app) share this build but
+//     are blocked at the host level by the canvas middleware sending
+//     an `X-Robots-Tag: noindex` header — robots.txt is per-host and
+//     this file's `host` field claims the apex as canonical.
+//
+// Note: `sitemap` is published via the sibling `sitemap.ts` route; we
+// reference it explicitly here so crawlers don't have to guess.
+const SITE_URL =
+  process.env.NEXT_PUBLIC_SITE_URL ?? "https://app.moleculesai.app";
+
+export default function robots(): MetadataRoute.Robots {
+  return {
+    rules: [
+      {
+        userAgent: "*",
+        allow: ["/", "/pricing", "/blog"],
+        // Authed app surface + API + transient checkout returns. The
+        // /orgs route boots the org-selector behind AuthGate; even
+        // though SSR returns markup, that markup is a login wall when
+        // hit by an unauthenticated crawler, so indexing it dilutes
+        // brand searches with a "Please sign in" snippet.
+        disallow: [
+          "/orgs",
+          "/orgs/",
+          "/api/",
+          "/cp/",
+          "/checkout/",
+        ],
+      },
+    ],
+    sitemap: `${SITE_URL}/sitemap.xml`,
+    host: SITE_URL,
+  };
+}
@@ -0,0 +1,42 @@
+import type { MetadataRoute } from "next";
+
+// Marketing-launch SEO (mc#1486). App-Router sitemap convention: this
+// file is served as `/sitemap.xml` and enumerates the public marketing
+// surface for search crawlers + AI training pipelines.
+//
+// Scope deliberately narrow:
+//   - Apex landing, pricing, and the (currently single) blog post.
+//   - Authed app routes are excluded — they're disallowed in robots.ts
+//     and would appear as "Please sign in" wall to a crawler.
+//
+// `lastModified` uses a build-time timestamp rather than per-route
+// fs.stat so the same value applies regardless of where the build
+// runs (Vercel/Railway/local). When we add CMS-backed blog content,
+// swap to a per-entry timestamp from the source-of-truth metadata.
+const SITE_URL =
+  process.env.NEXT_PUBLIC_SITE_URL ?? "https://app.moleculesai.app";
+
+const BUILD_DATE = new Date();
+
+export default function sitemap(): MetadataRoute.Sitemap {
+  return [
+    {
+      url: `${SITE_URL}/`,
+      lastModified: BUILD_DATE,
+      changeFrequency: "weekly",
+      priority: 1.0,
+    },
+    {
+      url: `${SITE_URL}/pricing`,
+      lastModified: BUILD_DATE,
+      changeFrequency: "weekly",
+      priority: 0.9,
+    },
+    {
+      url: `${SITE_URL}/blog/2026-04-20-chrome-devtools-mcp`,
+      lastModified: new Date("2026-04-20"),
+      changeFrequency: "monthly",
+      priority: 0.6,
+    },
+  ];
+}
@@ -132,7 +132,7 @@ export function AuditTrailPanel({ workspaceId }: Props) {

  if (loading) {
    return (
-      <div className="flex items-center justify-center h-32">
+      <div role="status" aria-live="polite" className="flex items-center justify-center h-32">
        <span className="text-xs text-ink-mid">Loading audit trail…</span>
      </div>
    );
@@ -133,13 +133,13 @@ export function ConversationTraceModal({ open, workspaceId: _workspaceId, onClos
            {/* Timeline */}
            <div className="flex-1 overflow-y-auto px-5 py-4">
              {loading && (
-                <div className="text-xs text-ink-mid text-center py-8">
+                <div role="status" aria-live="polite" className="text-xs text-ink-mid text-center py-8">
                  Loading trace from all workspaces...
                </div>
              )}

              {!loading && entries.length === 0 && (
-                <div className="text-xs text-ink-mid text-center py-8">
+                <div role="status" aria-live="polite" className="text-xs text-ink-mid text-center py-8">
                  No activity found
                </div>
              )}
@@ -105,7 +105,7 @@ export function EmptyState() {

        {/* Template grid */}
        {loading ? (
-          <div className="flex items-center justify-center gap-2 text-xs text-ink-mid py-4">
+          <div role="status" aria-live="polite" className="flex items-center justify-center gap-2 text-xs text-ink-mid py-4">
            <Spinner />
            Loading templates...
          </div>
@@ -15,7 +15,7 @@
 //     ($AGENT_URL). They ARE NOT filled in server-side because the
 //     server doesn't know where the operator's agent will live.

-import { useCallback, useState } from "react";
+import { useCallback, useRef, useState } from "react";
 import * as Dialog from "@radix-ui/react-dialog";

 type Tab = "python" | "curl" | "claude" | "mcp" | "hermes" | "codex" | "openclaw" | "kimi" | "fields";
@@ -84,6 +84,33 @@ export function ExternalConnectModal({ info, onClose }: Props) {
    : "python";
  const [tab, setTab] = useState<Tab>(initialTab);
  const [copiedKey, setCopiedKey] = useState<string | null>(null);
+  const tabRefs = useRef<Map<Tab, HTMLButtonElement | null>>(new Map());
+
+  const handleTabKeyDown = useCallback(
+    (e: React.KeyboardEvent<HTMLButtonElement>, current: Tab, tabs: Tab[]) => {
+      const idx = tabs.indexOf(current);
+      if (e.key === "ArrowRight" || e.key === "ArrowDown") {
+        e.preventDefault();
+        const next = tabs[(idx + 1) % tabs.length];
+        setTab(next);
+        tabRefs.current.get(next)?.focus();
+      } else if (e.key === "ArrowLeft" || e.key === "ArrowUp") {
+        e.preventDefault();
+        const prev = tabs[(idx - 1 + tabs.length) % tabs.length];
+        setTab(prev);
+        tabRefs.current.get(prev)?.focus();
+      } else if (e.key === "Home") {
+        e.preventDefault();
+        setTab(tabs[0]);
+        tabRefs.current.get(tabs[0])?.focus();
+      } else if (e.key === "End") {
+        e.preventDefault();
+        setTab(tabs[tabs.length - 1]);
+        tabRefs.current.get(tabs[tabs.length - 1])?.focus();
+      }
+    },
+    [],
+  );

  const copy = useCallback(async (value: string, key: string) => {
    try {
@@ -160,6 +187,19 @@ export function ExternalConnectModal({ info, onClose }: Props) {
    `MOLECULE_WORKSPACE_TOKEN=${info.auth_token}`,
  );

+  // Build the tab list once so both the tab bar and keyboard handler
+  // share the same ordered array. Computed here (after all filled* vars)
+  // so TypeScript's block-scoping analysis can reach them.
+  const tabList: Tab[] = [];
+  if (filledUniversalMcp) tabList.push("mcp");
+  tabList.push("python");
+  if (filledChannel) tabList.push("claude");
+  if (filledHermes) tabList.push("hermes");
+  if (filledCodex) tabList.push("codex");
+  if (filledOpenClaw) tabList.push("openclaw");
+  if (filledKimi) tabList.push("kimi");
+  tabList.push("curl", "fields");
+
  return (
    <Dialog.Root open onOpenChange={(o) => !o && onClose()}>
      <Dialog.Portal>
@@ -180,34 +220,18 @@ export function ExternalConnectModal({ info, onClose }: Props) {
            aria-label="Connection snippet format"
            className="mt-4 flex gap-1 border-b border-line"
          >
-            {(() => {
-              // Build the tab order dynamically. Claude Code first
-              // (when offered) since it's the simplest setup; Python
-              // SDK second (full register+heartbeat+inbound); Universal
-              // MCP third (any MCP-aware runtime, outbound-only); curl
-              // for one-shot register; Fields for raw values.
-              // Tab order: Universal MCP first (default, runtime-
-              // agnostic primitives), then runtime-specific channel/
-              // SDK tabs, then curl + Fields. Each runtime tab only
-              // appears when the platform supplies the snippet — no
-              // dead "tab missing snippet" UX.
-              const tabs: Tab[] = [];
-              if (filledUniversalMcp) tabs.push("mcp");
-              tabs.push("python");
-              if (filledChannel) tabs.push("claude");
-              if (filledHermes) tabs.push("hermes");
-              if (filledCodex) tabs.push("codex");
-              if (filledOpenClaw) tabs.push("openclaw");
-              if (filledKimi) tabs.push("kimi");
-              tabs.push("curl", "fields");
-              return tabs;
-            })().map((t) => (
+            {tabList.map((t) => (
              <button
                key={t}
                type="button"
                role="tab"
+                id={`tab-${t}`}
                aria-selected={tab === t}
+                aria-controls={`panel-${t}`}
+                tabIndex={tab === t ? 0 : -1}
+                ref={(el) => { tabRefs.current.set(t, el); }}
                onClick={() => setTab(t)}
+                onKeyDown={(e) => handleTabKeyDown(e, t, tabList)}
                className={`px-3 py-2 text-sm border-b-2 -mb-px transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-surface ${
                  tab === t
                    ? "border-accent text-ink"
@@ -235,18 +259,39 @@ export function ExternalConnectModal({ info, onClose }: Props) {
            ))}
          </div>

-          {/* Snippet area */}
-          <div className="mt-3">
-            {tab === "claude" && filledChannel && (
-              <SnippetBlock
-                value={filledChannel}
-                label="Claude Code channel — polls workspace's A2A; no tunnel needed"
-                copyKey="claude"
-                copied={copiedKey === "claude"}
-                onCopy={() => copy(filledChannel, "claude")}
-              />
-            )}
-            {tab === "python" && (
+          {/* Snippet area — all panels always in the DOM so aria-controls
+              targets are stable. Hidden panels use aria-hidden so screen
+              readers skip them; active panel uses role=tabpanel with
+              aria-labelledby pointing to the tab button. */}
+          <div className="mt-3" data-testid="snippet-panels">
+            {/* Claude Code tab */}
+            <div
+              id="panel-claude"
+              data-testid="panel-claude"
+              role="tabpanel"
+              aria-labelledby="tab-claude"
+              hidden={tab !== "claude" || !filledChannel}
+              className={tab === "claude" && filledChannel ? "" : "hidden"}
+            >
+              {filledChannel && (
+                <SnippetBlock
+                  value={filledChannel}
+                  label="Claude Code channel — polls workspace's A2A; no tunnel needed"
+                  copyKey="claude"
+                  copied={copiedKey === "claude"}
+                  onCopy={() => copy(filledChannel, "claude")}
+                />
+              )}
+            </div>
+            {/* Python SDK tab */}
+            <div
+              id="panel-python"
+              data-testid="panel-python"
+              role="tabpanel"
+              aria-labelledby="tab-python"
+              hidden={tab !== "python"}
+              className={tab === "python" ? "" : "hidden"}
+            >
              <SnippetBlock
                value={filledPython}
                label="Python SDK — includes heartbeat loop (push-mode, needs public URL)"
@@ -254,8 +299,16 @@ export function ExternalConnectModal({ info, onClose }: Props) {
                copied={copiedKey === "python"}
                onCopy={() => copy(filledPython, "python")}
              />
-            )}
-            {tab === "curl" && (
+            </div>
+            {/* curl tab */}
+            <div
+              id="panel-curl"
+              data-testid="panel-curl"
+              role="tabpanel"
+              aria-labelledby="tab-curl"
+              hidden={tab !== "curl"}
+              className={tab === "curl" ? "" : "hidden"}
+            >
              <SnippetBlock
                value={filledCurl}
                label="curl — one-shot register only (no heartbeat)"
@@ -263,53 +316,111 @@ export function ExternalConnectModal({ info, onClose }: Props) {
                copied={copiedKey === "curl"}
                onCopy={() => copy(filledCurl, "curl")}
              />
-            )}
-            {tab === "mcp" && filledUniversalMcp && (
-              <SnippetBlock
-                value={filledUniversalMcp}
-                label="Universal MCP — standalone register + heartbeat + tools for any MCP-aware runtime (Claude Code, hermes, codex). Pair with Python or Claude Code tab if you need inbound A2A delivery."
-                copyKey="mcp"
-                copied={copiedKey === "mcp"}
-                onCopy={() => copy(filledUniversalMcp, "mcp")}
-              />
-            )}
-            {tab === "hermes" && filledHermes && (
-              <SnippetBlock
-                value={filledHermes}
-                label="Hermes channel — bridges this workspace's A2A traffic into your hermes-agent session as platform messages (push parity with Claude Code). Long-poll based; no tunnel needed."
-                copyKey="hermes"
-                copied={copiedKey === "hermes"}
-                onCopy={() => copy(filledHermes, "hermes")}
-              />
-            )}
-            {tab === "codex" && filledCodex && (
-              <SnippetBlock
-                value={filledCodex}
-                label="Codex MCP config — wires the molecule MCP server into ~/.codex/config.toml. Outbound tools today; inbound A2A push needs the Python SDK tab paired in (codex's MCP runtime doesn't route arbitrary notifications/* yet)."
-                copyKey="codex"
-                copied={copiedKey === "codex"}
-                onCopy={() => copy(filledCodex, "codex")}
-              />
-            )}
-            {tab === "openclaw" && filledOpenClaw && (
-              <SnippetBlock
-                value={filledOpenClaw}
-                label="OpenClaw MCP config — wires the molecule MCP server via openclaw mcp set + starts the gateway on loopback. Outbound tools today; inbound A2A push on an external openclaw needs the Python SDK tab paired in (a sessions.steer bridge daemon is future work)."
-                copyKey="openclaw"
-                copied={copiedKey === "openclaw"}
-                onCopy={() => copy(filledOpenClaw, "openclaw")}
-              />
-            )}
-            {tab === "kimi" && filledKimi && (
-              <SnippetBlock
-                value={filledKimi}
-                label="Kimi CLI — self-contained Python bridge. Registers, heartbeats, polls for canvas messages, and echoes replies back. NAT-safe (no public URL). Run in a background terminal or via launchd."
-                copyKey="kimi"
-                copied={copiedKey === "kimi"}
-                onCopy={() => copy(filledKimi, "kimi")}
-              />
-            )}
-            {tab === "fields" && (
+            </div>
+            {/* Universal MCP tab */}
+            <div
+              id="panel-mcp"
+              data-testid="panel-mcp"
+              role="tabpanel"
+              aria-labelledby="tab-mcp"
+              hidden={tab !== "mcp" || !filledUniversalMcp}
+              className={tab === "mcp" && filledUniversalMcp ? "" : "hidden"}
+            >
+              {filledUniversalMcp && (
+                <SnippetBlock
+                  value={filledUniversalMcp}
+                  label="Universal MCP — standalone register + heartbeat + tools for any MCP-aware runtime (Claude Code, hermes, codex). Pair with Python or Claude Code tab if you need inbound A2A delivery."
+                  copyKey="mcp"
+                  copied={copiedKey === "mcp"}
+                  onCopy={() => copy(filledUniversalMcp, "mcp")}
+                />
+              )}
+            </div>
+            {/* Hermes tab */}
+            <div
+              id="panel-hermes"
+              data-testid="panel-hermes"
+              role="tabpanel"
+              aria-labelledby="tab-hermes"
+              hidden={tab !== "hermes" || !filledHermes}
+              className={tab === "hermes" && filledHermes ? "" : "hidden"}
+            >
+              {filledHermes && (
+                <SnippetBlock
+                  value={filledHermes}
+                  label="Hermes channel — bridges this workspace's A2A traffic into your hermes-agent session as platform messages (push parity with Claude Code). Long-poll based; no tunnel needed."
+                  copyKey="hermes"
+                  copied={copiedKey === "hermes"}
+                  onCopy={() => copy(filledHermes, "hermes")}
+                />
+              )}
+            </div>
+            {/* Codex tab */}
+            <div
+              id="panel-codex"
+              data-testid="panel-codex"
+              role="tabpanel"
+              aria-labelledby="tab-codex"
+              hidden={tab !== "codex" || !filledCodex}
+              className={tab === "codex" && filledCodex ? "" : "hidden"}
+            >
+              {filledCodex && (
+                <SnippetBlock
+                  value={filledCodex}
+                  label="Codex MCP config — wires the molecule MCP server into ~/.codex/config.toml. Outbound tools today; inbound A2A push needs the Python SDK tab paired in (codex's MCP runtime doesn't route arbitrary notifications/* yet)."
+                  copyKey="codex"
+                  copied={copiedKey === "codex"}
+                  onCopy={() => copy(filledCodex, "codex")}
+                />
+              )}
+            </div>
+            {/* OpenClaw tab */}
+            <div
+              id="panel-openclaw"
+              data-testid="panel-openclaw"
+              role="tabpanel"
+              aria-labelledby="tab-openclaw"
+              hidden={tab !== "openclaw" || !filledOpenClaw}
+              className={tab === "openclaw" && filledOpenClaw ? "" : "hidden"}
+            >
+              {filledOpenClaw && (
+                <SnippetBlock
+                  value={filledOpenClaw}
+                  label="OpenClaw MCP config — wires the molecule MCP server via openclaw mcp set + starts the gateway on loopback. Outbound tools today; inbound A2A push on an external openclaw needs the Python SDK tab paired in (a sessions.steer bridge daemon is future work)."
+                  copyKey="openclaw"
+                  copied={copiedKey === "openclaw"}
+                  onCopy={() => copy(filledOpenClaw, "openclaw")}
+                />
+              )}
+            </div>
+            {/* Kimi tab */}
+            <div
+              id="panel-kimi"
+              data-testid="panel-kimi"
+              role="tabpanel"
+              aria-labelledby="tab-kimi"
+              hidden={tab !== "kimi" || !filledKimi}
+              className={tab === "kimi" && filledKimi ? "" : "hidden"}
+            >
+              {filledKimi && (
+                <SnippetBlock
+                  value={filledKimi}
+                  label="Kimi CLI — self-contained Python bridge. Registers, heartbeats, polls for canvas messages, and echoes replies back. NAT-safe (no public URL). Run in a background terminal or via launchd."
+                  copyKey="kimi"
+                  copied={copiedKey === "kimi"}
+                  onCopy={() => copy(filledKimi, "kimi")}
+                />
+              )}
+            </div>
+            {/* Fields tab */}
+            <div
+              id="panel-fields"
+              data-testid="panel-fields"
+              role="tabpanel"
+              aria-labelledby="tab-fields"
+              hidden={tab !== "fields"}
+              className={tab === "fields" ? "" : "hidden"}
+            >
              <div className="space-y-2">
                <Field label="workspace_id" value={info.workspace_id} onCopy={() => copy(info.workspace_id, "wsid")} copied={copiedKey === "wsid"} />
                <Field label="platform_url" value={info.platform_url} onCopy={() => copy(info.platform_url, "url")} copied={copiedKey === "url"} />
@@ -323,7 +434,7 @@ export function ExternalConnectModal({ info, onClose }: Props) {
                <Field label="registry_endpoint" value={info.registry_endpoint} onCopy={() => copy(info.registry_endpoint, "reg")} copied={copiedKey === "reg"} />
                <Field label="heartbeat_endpoint" value={info.heartbeat_endpoint} onCopy={() => copy(info.heartbeat_endpoint, "hb")} copied={copiedKey === "hb"} />
              </div>
-            )}
+            </div>
          </div>

          <div className="mt-5 flex justify-end gap-2">
@@ -440,6 +440,7 @@ function ProviderPickerModal({
                      onChange={(e) => updateEntry(index, { value: e.target.value.trimStart() })}
                      placeholder={entry.key.includes("API_KEY") ? "sk-..." : "Enter value"}
                      type="password"
+                      aria-label={`Value for ${entry.key}`}
                      ref={index === 0 ? firstInputRef : undefined}
                      onKeyDown={(e) => {
                        if (e.key === "Enter" && entry.value.trim()) {
@@ -459,7 +460,7 @@ function ProviderPickerModal({
                )}

                {entry.error && (
-                  <div className="mt-1.5 text-[10px] text-bad">{entry.error}</div>
+                  <div role="alert" aria-live="assertive" className="mt-1.5 text-[10px] text-bad">{entry.error}</div>
                )}
              </div>
            ))}
@@ -694,6 +695,7 @@ function AllKeysModal({
                    onChange={(e) => updateEntry(index, { value: e.target.value.trimStart() })}
                    placeholder={entry.key.includes("API_KEY") ? "sk-..." : "Enter value"}
                    type="password"
+                    aria-label={`Value for ${entry.key}`}
                    autoFocus={index === 0}
                    onKeyDown={(e) => {
                      if (e.key === "Enter" && entry.value.trim()) {
@@ -718,7 +720,7 @@ function AllKeysModal({
          ))}

          {globalError && (
-            <div className="px-3 py-2 bg-red-950/40 border border-red-800/50 rounded-lg text-[11px] text-bad">
+            <div role="alert" aria-live="assertive" className="px-3 py-2 bg-red-950/40 border border-red-800/50 rounded-lg text-[11px] text-bad">
              {globalError}
            </div>
          )}
@@ -68,14 +68,103 @@ export function Toolbar() {
    return c;
  }, [nodes]);

+  /**
+   * Stop All - task #377 fix.
+   *
+   * BEFORE this PR: directly POSTed `/workspaces/:id/restart`, which tears
+   * the container down and back up. That kills in-flight tool subprocesses
+   * (e.g. `bash -c 'sleep 600'`) but is heavy and discards any in-progress
+   * agent state. It also bypasses the runtime-side fast cancel path (task
+   * #377 PR#40 in template-claude-code) - meaning flipping
+   * `MOLECULE_STOP_PROPAGATE=true` would produce zero canary signal because
+   * nothing ever invokes `executor.cancel()` in production.
+   *
+   * AFTER this PR (two-phase polite cancel):
+   *
+   * 1. POST `tasks/cancel` (A2A JSON-RPC) to each active workspace's
+   *    `/workspaces/:id/a2a` proxy. The platform proxies the envelope to
+   *    the workspace runtime; the a2a-sdk framework dispatches `tasks/cancel`
+   *    to `AgentExecutor.cancel()` (a2a-sdk 1.0.3
+   *    `a2a/compat/v0_3/types.py` line 1125 pins the wire literal as
+   *    `Literal["tasks/cancel"]`; A2A protocol spec section 9.4.5 maps the
+   *    abstract `CancelTask` operation to that wire string). The runtime's
+   *    executor cancel path signals the CLI subprocess group with
+   *    SIGTERM/grace/SIGKILL (template-claude-code PR#40 `stop_propagate.py`).
+   *
+   * 2. Poll the canvas store (the platform pushes `TASK_UPDATED` over WS
+   *    on `active_tasks` changes - `canvas-events.ts` line 400) for up to
+   *    `STOP_ALL_DRAIN_TIMEOUT_MS`. A workspace whose `activeTasks` drops
+   *    to 0 is considered drained and is NOT restarted.
+   *
+   * 3. For any workspace that DID NOT drain inside the timeout - runtime
+   *    is on an old image without the cancel path, or the cancel
+   *    propagation is stuck - fall back to the original heavy
+   *    `/workspaces/:id/restart`. The original behavior is preserved as a
+   *    floor so a stuck workspace still gets stopped; the polite path is
+   *    a fast top-up that lets well-behaved workspaces cancel without
+   *    losing context.
+   *
+   * The polite-cancel envelope mirrors `ScheduleTab.handleRunNow` (line 168)
+   * which is the only other place in canvas that POSTs `/workspaces/:id/a2a`
+   * directly. Method string `tasks/cancel` and empty `params` match the
+   * a2a-sdk shape verified above. The proxy adds `jsonrpc:"2.0"` and `id`
+   * via `normalizeA2APayload` server-side, so the canvas envelope omits them.
+   */
  const stopAll = useCallback(async () => {
    setStopping(true);
    const active = nodes.filter((n) => (n.data.activeTasks as number) > 0);
+    const activeIds = active.map((n) => n.id);
+
+    // Phase 1 - polite cancel on every active workspace in parallel.
+    // Errors are swallowed (same shape as the pre-fix /restart
+    // Promise.all): a 4xx/5xx on tasks/cancel just means we fall through
+    // to /restart for that workspace below.
    await Promise.all(
-      active.map((n) =>
-        api.post(`/workspaces/${n.id}/restart`).catch(() => {})
+      activeIds.map((id) =>
+        api
+          .post(`/workspaces/${id}/a2a`, {
+            method: "tasks/cancel",
+            params: {},
+          })
+          .catch(() => {})
      )
    );
+
+    // Phase 2 - poll the store for activeTasks reaching 0, with a hard
+    // timeout. STOP_ALL_DRAIN_TIMEOUT_MS is sized to cover the runtime's
+    // own SIGTERM-grace (5s in template-claude-code stop_propagate.py
+    // `_SIGTERM_GRACE_S`) plus a small WS round-trip buffer for the
+    // TASK_UPDATED push. STOP_ALL_POLL_INTERVAL_MS keeps the poll cheap
+    // (no animation jitter, no busy-wait).
+    const STOP_ALL_DRAIN_TIMEOUT_MS = 8000;
+    const STOP_ALL_POLL_INTERVAL_MS = 250;
+    const deadline = Date.now() + STOP_ALL_DRAIN_TIMEOUT_MS;
+    let undrained = new Set(activeIds);
+    while (undrained.size > 0 && Date.now() < deadline) {
+      await new Promise((r) => setTimeout(r, STOP_ALL_POLL_INTERVAL_MS));
+      const fresh = useCanvasStore.getState().nodes;
+      const stillActive = new Set<string>();
+      for (const id of undrained) {
+        const n = fresh.find((x) => x.id === id);
+        // Missing node (workspace deleted mid-cancel) is treated as
+        // drained - there's nothing left to restart and reporting it as
+        // "still running" would be a lie.
+        if (n && (n.data.activeTasks as number) > 0) stillActive.add(id);
+      }
+      undrained = stillActive;
+    }
+
+    // Phase 3 - hard-restart anything that did not drain. This is the
+    // same call shape as the pre-fix Stop All, so behavior is strictly a
+    // superset: undrained workspaces still get the heavy stop, drained
+    // ones are spared.
+    if (undrained.size > 0) {
+      await Promise.all(
+        Array.from(undrained).map((id) =>
+          api.post(`/workspaces/${id}/restart`).catch(() => {})
+        )
+      );
+    }
    setStopping(false);
  }, [nodes]);

@@ -71,7 +71,7 @@ export function WorkspaceUsage({ workspaceId }: WorkspaceUsageProps) {
            <SkeletonRow />
          </>
        ) : error ? (
-          <p className="text-xs text-bad" data-testid="usage-error">
+          <p role="alert" aria-live="assertive" className="text-xs text-bad" data-testid="usage-error">
            {error}
          </p>
        ) : metrics ? (
@@ -131,7 +131,9 @@ describe("ExternalConnectModal — tab switching", () => {
  it("switches to the Python SDK tab and shows the snippet with stamped token", () => {
    renderAndFlush(defaultInfo);
    fireEvent.click(screen.getByRole("tab", { name: /python sdk/i }));
-    const preEl = document.querySelector("pre");
+    // Query within the python panel so we get the right pre (not the first in DOM).
+    const pythonPanel = document.querySelector("[data-testid='panel-python']");
+    const preEl = pythonPanel?.querySelector("pre");
    expect(preEl?.textContent).toContain("AUTH_TOKEN");
    // The placeholder is replaced with the real auth token
    expect(preEl?.textContent).toContain("secret-auth-token-abc");
@@ -140,7 +142,9 @@ describe("ExternalConnectModal — tab switching", () => {
  it("switches to the curl tab and shows the snippet with stamped token", () => {
    renderAndFlush(defaultInfo);
    fireEvent.click(screen.getByRole("tab", { name: /curl/i }));
-    const preEl = document.querySelector("pre");
+    // Query within the curl panel so we get the right pre (not the first in DOM).
+    const curlPanel = document.querySelector("[data-testid='panel-curl']");
+    const preEl = curlPanel?.querySelector("pre");
    expect(preEl?.textContent).toContain("curl");
    expect(preEl?.textContent).toContain("secret-auth-token-abc");
  });
@@ -148,9 +152,11 @@ describe("ExternalConnectModal — tab switching", () => {
  it("switches to the Fields tab and shows raw values", () => {
    renderAndFlush(defaultInfo);
    fireEvent.click(screen.getByRole("tab", { name: /fields/i }));
-    expect(screen.getByText("ws-123")).toBeTruthy();
-    expect(screen.getByText("https://app.example.com")).toBeTruthy();
-    expect(screen.getByText("secret-auth-token-abc")).toBeTruthy();
+    // Query within the fields panel for specific values.
+    const fieldsPanel = document.querySelector("[data-testid='panel-fields']");
+    expect(fieldsPanel?.textContent).toContain("ws-123");
+    expect(fieldsPanel?.textContent).toContain("https://app.example.com");
+    expect(fieldsPanel?.textContent).toContain("secret-auth-token-abc");
  });

  it("hides the Hermes tab when hermes_channel_snippet is absent", () => {
@@ -168,7 +174,8 @@ describe("ExternalConnectModal — snippet token stamping", () => {
  it("stamps the real auth_token into the Python snippet instead of the placeholder", () => {
    renderAndFlush(defaultInfo);
    fireEvent.click(screen.getByRole("tab", { name: /python sdk/i }));
-    const preEl = document.querySelector("pre");
+    const pythonPanel = document.querySelector("[data-testid='panel-python']");
+    const preEl = pythonPanel?.querySelector("pre");
    expect(preEl?.textContent).not.toContain("<paste from create response>");
    expect(preEl?.textContent).toContain("secret-auth-token-abc");
  });
@@ -176,7 +183,8 @@ describe("ExternalConnectModal — snippet token stamping", () => {
  it("stamps the real auth_token into the curl snippet", () => {
    renderAndFlush(defaultInfo);
    fireEvent.click(screen.getByRole("tab", { name: /curl/i }));
-    const preEl = document.querySelector("pre");
+    const curlPanel = document.querySelector("[data-testid='panel-curl']");
+    const preEl = curlPanel?.querySelector("pre");
    // curl template uses WORKSPACE_AUTH_TOKEN placeholder, not the generic one
    expect(preEl?.textContent).toContain("secret-auth-token-abc");
  });
@@ -184,7 +192,8 @@ describe("ExternalConnectModal — snippet token stamping", () => {
  it("stamps the real auth_token into the Universal MCP snippet", () => {
    renderAndFlush(defaultInfo);
    // Default tab is Universal MCP
-    const preEl = document.querySelector("pre");
+    const mcpPanel = document.querySelector("[data-testid='panel-mcp']");
+    const preEl = mcpPanel?.querySelector("pre");
    expect(preEl?.textContent).toContain("secret-auth-token-abc");
    expect(preEl?.textContent).not.toContain("<paste from create response>");
  });
@@ -193,8 +202,10 @@ describe("ExternalConnectModal — snippet token stamping", () => {
 describe("ExternalConnectModal — copy functionality", () => {
  it("calls navigator.clipboard.writeText with the snippet text", () => {
    renderAndFlush(defaultInfo);
-    // Default tab is Universal MCP
-    fireEvent.click(screen.getByRole("button", { name: /^copy$/i }));
+    // Default tab is Universal MCP — query the copy button within the mcp panel.
+    const mcpPanel = document.querySelector("[data-testid='panel-mcp']");
+    const copyBtn = mcpPanel?.querySelector("button");
+    if (copyBtn) fireEvent.click(copyBtn);
    expect(clipboardWriteText).toHaveBeenCalledWith(
      expect.stringContaining("secret-auth-token-abc"),
    );
@@ -227,7 +238,8 @@ describe("ExternalConnectModal — missing optional fields", () => {
    };
    renderAndFlush(minimalInfo);
    fireEvent.click(screen.getByRole("tab", { name: /fields/i }));
-    expect(screen.getByText("(missing)")).toBeTruthy();
+    const fieldsPanel = document.querySelector("[data-testid='panel-fields']");
+    expect(fieldsPanel?.textContent).toContain("(missing)");
  });

  it("hides the Hermes tab when hermes_channel_snippet is absent", () => {
@@ -11,13 +11,21 @@ import { render, screen, fireEvent, cleanup, act } from "@testing-library/react"
 import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
 import { TestConnectionButton } from "../ui/TestConnectionButton";
 import type { SecretGroup } from "@/types/secrets";
-import { validateSecret } from "@/lib/api/secrets";
+import { validateSecret, ApiError } from "@/lib/api/secrets";

 // ─── Mock validateSecret ──────────────────────────────────────────────────────
 // vi.mock is hoisted, so validateSecret (imported above) refers to the mocked
 // namespace value once vi.mock runs. Use vi.mocked() to access it in tests.
 vi.mock("@/lib/api/secrets", () => ({
  validateSecret: vi.fn(),
+  ApiError: class ApiError extends Error {
+    status: number;
+    constructor(status: number, message: string) {
+      super(message);
+      this.name = "ApiError";
+      this.status = status;
+    }
+  },
 }));

 // SecretGroup is a string literal type: 'github' | 'anthropic' | 'openrouter' | 'custom'
@@ -102,7 +110,7 @@ describe("TestConnectionButton — state machine", () => {
    expect(screen.getByText("Permission denied")).toBeTruthy();
  });

-  it("shows generic error message on unexpected exception", async () => {
+  it("shows a connectivity message on a genuine network exception", async () => {
    vi.mocked(validateSecret).mockRejectedValue(new Error("timeout"));
    render(<TestConnectionButton provider={toGroup("anthropic")} secretValue="sk-..." />);

@@ -110,8 +118,23 @@ describe("TestConnectionButton — state machine", () => {
    await act(async () => { /* flush */ });

    expect(screen.getByRole("alert")).toBeTruthy();
-    // The error detail is hardcoded to "Connection timed out. Service may be down."
-    expect(document.body.querySelector('[role="alert"]')?.textContent).toMatch(/timed out/i);
+    // A real thrown network error → honest connectivity message (not a
+    // fabricated "service down"); see internal#492.
+    expect(document.body.querySelector('[role="alert"]')?.textContent).toMatch(
+      /could not reach the validation service/i,
+    );
+  });
+
+  it("does not claim a timeout when the validate endpoint 404s (internal#492)", async () => {
+    vi.mocked(validateSecret).mockRejectedValue(new ApiError(404, "Not Found"));
+    render(<TestConnectionButton provider={toGroup("anthropic")} secretValue="sk-..." />);
+
+    fireEvent.click(screen.getByRole("button"));
+    await act(async () => { /* flush */ });
+
+    const alert = document.body.querySelector('[role="alert"]')?.textContent ?? "";
+    expect(alert).not.toMatch(/timed out/i);
+    expect(alert).toMatch(/not available/i);
  });
 });

@@ -131,14 +131,30 @@ const defaultStore = {
  batchDelete: vi.fn(() => Promise.resolve()),
 };

-vi.mock("@/store/canvas", () => ({
-  useCanvasStore: vi.fn((selector: (s: typeof defaultStore) => unknown) =>
+vi.mock("@/store/canvas", () => {
+  // useCanvasStore is used in two shapes:
+  //   1. As a hook: `useCanvasStore((s) => s.x)` — selector path.
+  //   2. As a static accessor: `useCanvasStore.getState().nodes` —
+  //      used by stopAll's drain-poll loop (task #377 Toolbar fix) and
+  //      restartAll's success-clear loop. Both read the LIVE
+  //      defaultStore object so tests that mutate `defaultStore.nodes`
+  //      mid-flight (e.g. simulating a TASK_UPDATED that drops
+  //      activeTasks to 0) see the update on the next poll tick.
+  const hook = vi.fn((selector: (s: typeof defaultStore) => unknown) =>
    selector(defaultStore)
-  ),
-}));
+  ) as unknown as ((selector: (s: typeof defaultStore) => unknown) => unknown) & {
+    getState: () => typeof defaultStore;
+  };
+  hook.getState = () => defaultStore;
+  return { useCanvasStore: hook };
+});

 // ── Component under test ───────────────────────────────────────────────────────
 import { Toolbar } from "../Toolbar";
+// Imported AFTER vi.mock("@/lib/api", ...) above (hoisted) so this
+// resolves to the mock module; gives the new task #377 tests a typed
+// handle on api.post without a CJS require() (Vitest runs ESM).
+import { api as mockedApi } from "@/lib/api";

 // ── Tests ─────────────────────────────────────────────────────────────────────

@@ -315,3 +331,157 @@ describe("Toolbar — ? shortcut opens shortcuts dialog", () => {
    expect(screen.queryByTestId("shortcuts-dialog")).toBeNull();
  });
 });
+
+// ── Toolbar — Stop All polite-cancel flow (task #377) ───────────────────────
+
+describe("Toolbar — Stop All polite cancel before restart (#377)", () => {
+  // `api` resolves to the top-level vi.mock factory's mocked `post`.
+  // We type-cast so TS allows mockReset/mockResolvedValue/mockImplementation
+  // calls without leaking the mock surface into the production type.
+  const api = mockedApi as unknown as { post: ReturnType<typeof vi.fn> };
+
+  /**
+   * Build a working set of two active workspaces so the assertions can
+   * distinguish per-id behavior (drained vs undrained) within one test.
+   */
+  const seedTwoActive = () => {
+    defaultStore.nodes = toStoreNodes(makeNodes(["online", "online"], [2, 2]));
+  };
+
+  /**
+   * Drive an async useCallback handler to completion. Vitest's fake
+   * timers don't see microtasks unless we yield between advances; the
+   * helper interleaves `vi.advanceTimersByTimeAsync` with macrotask
+   * yields so pending fetch resolutions and setTimeout callbacks both
+   * settle before the assertion runs.
+   */
+  const advanceUntilSettled = async (ms: number) => {
+    await vi.advanceTimersByTimeAsync(ms);
+    // One extra tick lets any chained .then() after a setTimeout
+    // resolution fire before the test moves on.
+    await Promise.resolve();
+  };
+
+  beforeEach(() => {
+    vi.useFakeTimers();
+    api.post.mockReset();
+  });
+
+  afterEach(() => {
+    vi.useRealTimers();
+  });
+
+  it("phase 1: issues tasks/cancel via /workspaces/:id/a2a BEFORE any /restart", async () => {
+    seedTwoActive();
+    // Hold both tasks/cancel responses open so the click handler is
+    // observably paused at phase 1. We don't actually need to resolve
+    // them for the order assertion — just inspect the call log.
+    let resolveCancels!: () => void;
+    const cancelGate = new Promise<void>((r) => { resolveCancels = r; });
+    api.post.mockImplementation(async (path: string) => {
+      if (path.endsWith("/a2a")) {
+        await cancelGate;
+      }
+      return undefined;
+    });
+
+    render(<Toolbar />);
+    const btn = screen.getByRole("button", { name: /stop all running tasks/i });
+    fireEvent.click(btn);
+
+    // Yield once so the click handler enters phase 1 and dispatches the
+    // two /a2a POSTs.
+    await Promise.resolve();
+    await Promise.resolve();
+
+    const a2aCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/a2a"));
+    const restartCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/restart"));
+    expect(a2aCalls.length).toBe(2);
+    expect(restartCalls.length).toBe(0);
+
+    // Each /a2a POST carries the canonical tasks/cancel envelope.
+    for (const call of a2aCalls) {
+      expect(call[1]).toEqual({ method: "tasks/cancel", params: {} });
+    }
+
+    // Release the gate so the test cleanup doesn't dangle.
+    resolveCancels();
+    await advanceUntilSettled(10_000);
+  });
+
+  it("phase 2: when activeTasks drains to 0 during the poll window, /restart is NOT called", async () => {
+    seedTwoActive();
+    api.post.mockResolvedValue(undefined);
+
+    render(<Toolbar />);
+    fireEvent.click(screen.getByRole("button", { name: /stop all running tasks/i }));
+
+    // Let phase 1 fire (the two tasks/cancel calls).
+    await Promise.resolve();
+    await Promise.resolve();
+
+    // Simulate the platform pushing TASK_UPDATED with active_tasks=0
+    // on both workspaces — emulate by mutating the store directly,
+    // which is what canvas-events.ts does in production.
+    defaultStore.nodes = toStoreNodes(makeNodes(["online", "online"], [0, 0]));
+
+    // Advance past the first poll interval (250ms) so the loop sees
+    // the drained store and exits early.
+    await advanceUntilSettled(400);
+    // Drain any remaining timers so the handler returns cleanly.
+    await advanceUntilSettled(10_000);
+
+    const restartCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/restart"));
+    expect(restartCalls.length).toBe(0);
+  });
+
+  it("phase 3: when activeTasks does NOT drain inside the timeout, falls through to /restart for each stuck workspace", async () => {
+    seedTwoActive();
+    api.post.mockResolvedValue(undefined);
+
+    render(<Toolbar />);
+    fireEvent.click(screen.getByRole("button", { name: /stop all running tasks/i }));
+
+    // Phase 1 dispatch.
+    await Promise.resolve();
+    await Promise.resolve();
+
+    // Do NOT drain — activeTasks stays at 2 for both. Advance past the
+    // 8000ms drain timeout plus a buffer so phase 3's /restart POSTs fire.
+    await advanceUntilSettled(9_000);
+    await advanceUntilSettled(1_000);
+
+    const a2aCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/a2a"));
+    const restartCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/restart"));
+    expect(a2aCalls.length).toBe(2);
+    expect(restartCalls.length).toBe(2);
+
+    // Order check: every /a2a call comes before every /restart call.
+    const lastA2AIdx = Math.max(
+      ...api.post.mock.calls.map((c, i) => (String(c[0]).endsWith("/a2a") ? i : -1))
+    );
+    const firstRestartIdx = Math.min(
+      ...api.post.mock.calls.map((c, i) => (String(c[0]).endsWith("/restart") ? i : Infinity))
+    );
+    expect(lastA2AIdx).toBeLessThan(firstRestartIdx);
+  });
+
+  it("phase 3 selective: drains only one of two workspaces — /restart is called only for the stuck one", async () => {
+    seedTwoActive();
+    api.post.mockResolvedValue(undefined);
+
+    render(<Toolbar />);
+    fireEvent.click(screen.getByRole("button", { name: /stop all running tasks/i }));
+
+    await Promise.resolve();
+    await Promise.resolve();
+
+    // ws-0 drains immediately, ws-1 stays stuck for the full timeout.
+    defaultStore.nodes = toStoreNodes(makeNodes(["online", "online"], [0, 2]));
+    await advanceUntilSettled(9_500);
+
+    const restartCalls = api.post.mock.calls.filter((c) => String(c[0]).endsWith("/restart"));
+    expect(restartCalls.length).toBe(1);
+    expect(restartCalls[0][0]).toBe("/workspaces/ws-1/restart");
+  });
+});
@@ -223,6 +223,7 @@ export function MobileCanvas({
            textTransform: "uppercase",
            fontWeight: 600,
          }}
+          className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
        >
          Reset
        </button>
@@ -2,8 +2,11 @@

 // 04 · Chat — message thread + composer + sub-tabs.
 // Wired to the same /workspaces/:id/a2a (method message/send) endpoint
-// that the desktop ChatTab uses, but with a slimmer surface: no
-// attachments, no A2A topology overlay, no conversation tracing.
+// that the desktop ChatTab uses. Render parity with desktop ChatTab is
+// achieved by reusing its renderers rather than forking a reduced
+// mobile path: the Agent Comms sub-tab mounts the same AgentCommsPanel,
+// and message attachments route through the same AttachmentPreview
+// dispatch the desktop My-Chat bubble uses (#231/#232).

 import { useEffect, useMemo, useRef, useState } from "react";
 import ReactMarkdown from "react-markdown";
@@ -16,6 +19,9 @@ import {
  useChatSend,
  useChatSocket,
 } from "@/components/tabs/chat/hooks";
+import { AgentCommsPanel } from "@/components/tabs/chat/AgentCommsPanel";
+import { AttachmentPreview } from "@/components/tabs/chat/AttachmentPreview";
+import { downloadChatFile } from "@/components/tabs/chat/uploads";

 import { toMobileAgent } from "./components";
 import { MOBILE_FONT_MONO, MOBILE_FONT_SANS, usePalette } from "./palette";
@@ -304,6 +310,17 @@ export function MobileChat({
  const removePendingFile = (index: number) =>
    setPendingFiles((prev) => prev.filter((_, i) => i !== index));

+  // Route attachment downloads through the same authenticated helper
+  // the desktop ChatTab uses (downloadChatFile) so platform-scheme
+  // URIs get a real Blob with auth headers instead of about:blank.
+  const downloadAttachment = (att: ChatAttachment) => {
+    downloadChatFile(agentId, att).catch(() => {
+      // AttachmentPreview's own error affordance covers the in-bubble
+      // failure state; matches ChatTab's behaviour of not double-
+      // reporting a download failure.
+    });
+  };
+
  const send = async () => {
    const text = draft.trim();
    if ((!text && pendingFiles.length === 0) || sending || !reachable) return;
@@ -339,6 +356,7 @@ export function MobileChat({
            type="button"
            onClick={onBack}
            aria-label="Back"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: 36,
              height: 36,
@@ -385,6 +403,7 @@ export function MobileChat({
          <button
            type="button"
            aria-label="More"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: 36,
              height: 36,
@@ -415,6 +434,7 @@ export function MobileChat({
                key={t.id}
                type="button"
                onClick={() => setTab(t.id)}
+                className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
                style={{
                  padding: "4px 0 8px",
                  border: "none",
@@ -433,7 +453,19 @@ export function MobileChat({
        </div>
      </div>

+      {/* Agent Comms — reuse the desktop AgentCommsPanel verbatim so
+          mobile renders the identical peer/A2A + delegation feed
+          (history GET + live socket events) instead of a placeholder
+          (#231). The panel owns its own scroll/load/error/empty
+          states, matching ChatTab's agent-comms tabpanel. */}
+      {tab === "a2a" && (
+        <div style={{ flex: 1, minHeight: 0, overflow: "hidden" }}>
+          <AgentCommsPanel workspaceId={agentId} />
+        </div>
+      )}
+
      {/* Messages */}
+      {tab === "my" && (
      <div
        ref={scrollRef}
        style={{
@@ -445,20 +477,8 @@ export function MobileChat({
          gap: 8,
        }}
      >
-        {tab === "a2a" && (
-          <div
-            style={{
-              padding: "20px 4px",
-              textAlign: "center",
-              color: p.text3,
-              fontSize: 13,
-            }}
-          >
-            Agent Comms — peer-to-peer A2A traffic surfaces in the Comms tab.
-          </div>
-        )}
        {tab === "my" && historyLoading && (
-          <div style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
+          <div role="status" aria-live="polite" style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
            Loading chat history…
          </div>
        )}
@@ -478,6 +498,8 @@ export function MobileChat({
              onClick={() => {
                loadInitial();
              }}
+              aria-label="Retry loading chat history"
+              className="focus:outline-none focus-visible:ring-2 focus-visible:ring-red-400"
              style={{
                padding: "6px 14px",
                borderRadius: 14,
@@ -493,7 +515,7 @@ export function MobileChat({
          </div>
        )}
        {tab === "my" && !historyLoading && !historyError && messages.length === 0 && (
-          <div style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
+          <div role="status" aria-live="polite" style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
            Send a message to start chatting.
          </div>
        )}
@@ -521,9 +543,31 @@ export function MobileChat({
                    overflowWrap: "anywhere",
                  }}
                >
-                  <MarkdownBubble dark={dark} accent={p.accent}>
-                    {m.content}
-                  </MarkdownBubble>
+                  {m.content && (
+                    <MarkdownBubble dark={dark} accent={p.accent}>
+                      {m.content}
+                    </MarkdownBubble>
+                  )}
+                  {m.attachments && m.attachments.length > 0 && (
+                    <div
+                      style={{
+                        display: "flex",
+                        flexWrap: "wrap",
+                        gap: 4,
+                        marginTop: m.content ? 6 : 0,
+                      }}
+                    >
+                      {m.attachments.map((att, i) => (
+                        <AttachmentPreview
+                          key={`${m.id}-${i}`}
+                          workspaceId={agentId}
+                          attachment={att}
+                          onDownload={downloadAttachment}
+                          tone={mine ? "user" : "agent"}
+                        />
+                      ))}
+                    </div>
+                  )}
                  <div
                    style={{
                      fontSize: 10,
@@ -554,7 +598,13 @@ export function MobileChat({
          </div>
        )}
      </div>
+      )}

+      {/* Footer ID + composer belong to My Chat only. The Agent Comms
+          tab is a read-only peer/A2A feed (parity with desktop
+          ChatTab, where the agent-comms tabpanel has no composer). */}
+      {tab === "my" && (
+      <>
      {/* Footer ID */}
      <div
        style={{
@@ -619,6 +669,7 @@ export function MobileChat({
                  type="button"
                  onClick={() => removePendingFile(i)}
                  aria-label={`Remove ${f.name}`}
+                  className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
                  style={{
                    border: "none",
                    background: "transparent",
@@ -659,6 +710,7 @@ export function MobileChat({
            onClick={() => fileInputRef.current?.click()}
            disabled={!reachable || sending || uploading}
            aria-label="Attach"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: 32,
              height: 32,
@@ -680,6 +732,7 @@ export function MobileChat({
            ref={composerRef}
            value={draft}
            onChange={(e) => setDraft(e.target.value)}
+            aria-label="Message"
            onKeyDown={(e) => {
              // Enter sends; Shift+Enter inserts a newline. Skip when the
              // IME is composing — pressing Enter to commit a Chinese/
@@ -703,7 +756,12 @@ export function MobileChat({
              border: "none",
              outline: "none",
              background: "transparent",
-              fontSize: 14.5,
+              // iOS Safari/PWA zooms the viewport when a focused textarea
+              // has a computed font-size below 16px. 14.5 triggers that
+              // focus-zoom; the page looks broken until the user pinches
+              // back (#224, same class as desktop #1434 / sibling #225).
+              // 16px is the minimum that keeps focus from zooming.
+              fontSize: 16,
              lineHeight: 1.4,
              color: p.text,
              padding: "6px 0",
@@ -719,12 +777,13 @@ export function MobileChat({
            onClick={send}
            disabled={(!draft.trim() && pendingFiles.length === 0) || !reachable || sending || uploading}
            aria-label="Send"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: 36,
              height: 36,
              borderRadius: 999,
              border: "none",
-              cursor: (draft.trim() || pendingFiles.length > 0) && !sending && !uploading ? "pointer" : "not-allowed",
+              cursor: (draft.trim() || pendingFiles.length === 0) && !sending && !uploading ? "pointer" : "not-allowed",
              flexShrink: 0,
              background:
                (draft.trim() || pendingFiles.length > 0) && reachable && !sending && !uploading
@@ -746,6 +805,8 @@ export function MobileChat({
          </button>
        </div>
      </div>
+      </>
+      )}
    </div>
  );
 }
@@ -231,6 +231,7 @@ export function MobileComms({ dark }: { dark: boolean }) {
                fontSize: 13,
                fontWeight: 500,
              }}
+              className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            >
              {o.label}
              <span
@@ -251,11 +252,11 @@ export function MobileComms({ dark }: { dark: boolean }) {

      <div style={{ padding: "0 14px", display: "flex", flexDirection: "column", gap: 8 }}>
        {loading && items.length === 0 ? (
-          <div style={{ padding: "30px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
+          <div role="status" aria-live="polite" style={{ padding: "30px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
            Loading recent comms…
          </div>
        ) : filtered.length === 0 ? (
-          <div style={{ padding: "30px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
+          <div role="status" aria-live="polite" style={{ padding: "30px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
            No A2A traffic yet.
          </div>
        ) : (
@@ -83,11 +83,12 @@ export function MobileDetail({
            type="button"
            onClick={onBack}
            aria-label="Back"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={iconButtonStyle(p, dark)}
          >
            {Icons.back({ size: 18 })}
          </button>
-          <button type="button" aria-label="More" style={iconButtonStyle(p, dark)}>
+          <button type="button" aria-label="More" className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900" style={iconButtonStyle(p, dark)}>
            {Icons.more({ size: 18 })}
          </button>
        </div>
@@ -183,6 +184,7 @@ export function MobileDetail({
              key={t.id}
              type="button"
              onClick={() => setTab(t.id)}
+              className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
              style={{
                padding: "8px 14px",
                borderRadius: 999,
@@ -215,6 +217,7 @@ export function MobileDetail({
          type="button"
          onClick={onChat}
          data-testid="mobile-chat-cta"
+          className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
          style={{
            width: "100%",
            height: 52,
@@ -416,6 +419,8 @@ function DetailActivity({ workspaceId, dark }: { workspaceId: string; dark: bool
  if (items === null) {
    return (
      <div
+        role="status"
+        aria-live="polite"
        style={{
          background: p.surface,
          borderRadius: 16,
@@ -200,6 +200,7 @@ export function MobileHome({
          justifyContent: "center",
          boxShadow: "0 8px 24px rgba(40,30,20,0.25), 0 2px 6px rgba(40,30,20,0.15)",
        }}
+        className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
      >
        {Icons.plus({ size: 22 })}
      </button>
@@ -92,6 +92,7 @@ export function MobileMe({
                    border: on ? `2px solid ${p.text}` : "2px solid transparent",
                    boxShadow: on ? `0 0 0 2px ${p.bg} inset` : "none",
                  }}
+                  className="focus:outline-none focus-visible:ring-2 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
                />
              );
            })}
@@ -184,6 +185,7 @@ function SegmentedRow({
              fontSize: 13,
              fontWeight: 600,
            }}
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
          >
            {o.label}
          </button>
@@ -148,6 +148,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
            type="button"
            onClick={onClose}
            aria-label="Close"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: 32,
              height: 32,
@@ -170,6 +171,8 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
        <div style={{ padding: "0 14px" }}>
          {loadingTemplates ? (
            <div
+              role="status"
+              aria-live="polite"
              style={{
                padding: "24px 8px",
                textAlign: "center",
@@ -214,6 +217,8 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
                      setTplId(t.id);
                      setTier(tCode);
                    }}
+                    aria-label={`Select template: ${t.name} (tier ${t.tier})`}
+                    className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
                    style={{
                      background: on
                        ? dark
@@ -302,6 +307,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
          <input
            value={name}
            onChange={(e) => setName(e.target.value)}
+            aria-label="Agent name"
            placeholder={tplId
              ? (templates.find((t) => t.id === tplId)?.name ?? "agent-name")
              : "agent-name"}
@@ -312,7 +318,12 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
              border: `0.5px solid ${p.border}`,
              borderRadius: 12,
              fontFamily: MOBILE_FONT_MONO,
-              fontSize: 13.5,
+              // iOS Safari/PWA zooms the viewport when a focused input has
+              // a computed font-size below 16px; the layout jumps and the
+              // page looks broken until the user pinches back (#224 / #225,
+              // same class as desktop #1434). 16px is the minimum that
+              // suppresses that focus-zoom.
+              fontSize: 16,
              color: p.text,
              outline: "none",
              boxSizing: "border-box",
@@ -330,6 +341,8 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
                key={t}
                type="button"
                onClick={() => setTier(t)}
+                aria-label={`Select tier ${t}: ${TIER_LABEL[t]}`}
+                className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
                style={{
                  flex: 1,
                  padding: "10px 8px",
@@ -377,6 +390,8 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
            type="button"
            onClick={handleSpawn}
            disabled={busy || !tplId || templates.length === 0}
+            aria-label="Spawn agent"
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              width: "100%",
              height: 52,
@@ -21,6 +21,14 @@ import { MobileChat } from "../MobileChat";
 vi.mock("@/lib/api");
 import { api } from "@/lib/api";

+// AgentCommsPanel (mounted by the Agent Comms sub-tab, #231) subscribes
+// to the global socket via useSocketEvent. Stub it to a no-op so the
+// panel mounts without the real ReconnectingSocket — the parity tests
+// only assert the panel renders (vs the old static placeholder).
+vi.mock("@/hooks/useSocketEvent", () => ({
+  useSocketEvent: vi.fn(),
+}));
+
 // ─── Mock store ───────────────────────────────────────────────────────────────

 const mockAgentId = "ws-chat-test";
@@ -155,6 +163,12 @@ beforeEach(() => {
  mockOnBack.mockClear();
  mockStoreState.nodes = [];
  mockStoreState.agentMessages = {};
+  // jsdom doesn't implement scrollIntoView. The Agent Comms tab now
+  // mounts AgentCommsPanel (#231), which scrolls its feed to bottom on
+  // arrival; a no-op stub keeps the panel from throwing under jsdom
+  // (same stub AgentCommsPanel's own render test installs).
+  Element.prototype.scrollIntoView =
+    vi.fn() as unknown as Element["scrollIntoView"];
  // Set up spies on the real api methods. Tests override these per-call.
  const getSpy = vi.spyOn(api, "get");
  const postSpy = vi.spyOn(api, "post");
@@ -249,6 +263,20 @@ describe("MobileChat — composer", () => {
    const sendBtn = container.querySelector('[aria-label="Send"]') as HTMLButtonElement;
    expect(sendBtn.disabled).toBe(true);
  });
+
+  // Regression #224: the composer textarea must render with font-size
+  // ≥ 16px. iOS Safari and PWAs auto-zoom the viewport when a focused
+  // input has a computed font-size below 16px — the layout jumps and
+  // the page looks broken until the user pinches back. Same class as
+  // desktop #1434 / sibling MobileSpawn #225.
+  it("composer textarea renders at font-size 16px or greater (iOS focus-zoom regression #224)", () => {
+    const { container } = renderChat(mockAgentId);
+    const textarea = container.querySelector("textarea") as HTMLTextAreaElement;
+    expect(textarea).toBeTruthy();
+    const fs = Number.parseFloat(textarea.style.fontSize);
+    expect(Number.isFinite(fs)).toBe(true);
+    expect(fs).toBeGreaterThanOrEqual(16);
+  });
 });

 // ─── Tabs ─────────────────────────────────────────────────────────────────────
@@ -474,3 +502,146 @@ describe("MobileChat — chat history", () => {
    expect(getSpy).toHaveBeenCalledTimes(2);
  });
 });
+
+// ─── #232 · Attachment render parity with desktop ChatTab ────────────────────
+//
+// Regression for the CTO-reported mobile bug: MobileChat used to render
+// only m.content (no attachment surface), so files sent/received in a
+// conversation were invisible on mobile while desktop showed them. The
+// fix routes m.attachments through the same AttachmentPreview the
+// desktop ChatTab bubble uses.
+
+describe("MobileChat — attachment render parity (#232)", () => {
+  beforeEach(() => {
+    mockStoreState.nodes = [onlineNode];
+  });
+
+  it("renders an attachment from a history message via AttachmentPreview", async () => {
+    const getSpy = vi.spyOn(api, "get");
+    // useChatHistory reads { messages, reached_end }.
+    getSpy.mockResolvedValueOnce({
+      messages: [
+        {
+          id: "m-att-1",
+          role: "agent",
+          content: "Here is the report",
+          attachments: [
+            {
+              name: "report.csv",
+              uri: "workspace://out/report.csv",
+              mimeType: "text/csv",
+              size: 2048,
+            },
+          ],
+          timestamp: new Date().toISOString(),
+        },
+      ],
+      reached_end: true,
+    });
+
+    let rr: ReturnType<typeof renderChat>;
+    await act(async () => {
+      rr = renderChat(mockAgentId);
+    });
+    const { container } = rr!;
+
+    // A non-image attachment renders the AttachmentChip download button
+    // with title="Download <name>" — same component the desktop bubble
+    // dispatches through AttachmentPreview.
+    await waitFor(() => {
+      const chip = container.querySelector('[title="Download report.csv"]');
+      expect(chip).toBeTruthy();
+    });
+    expect(container.textContent ?? "").toContain("report.csv");
+  });
+});
+
+// ─── #231 · Agent Comms (A2A/peer) render parity with desktop ChatTab ────────
+//
+// Regression for the CTO-reported mobile bug: the Agent Comms sub-tab
+// rendered a static placeholder string ("peer-to-peer A2A traffic
+// surfaces in the Comms tab") instead of the real feed. The fix mounts
+// the same AgentCommsPanel the desktop ChatTab agent-comms tabpanel
+// uses, so peer/A2A + delegation activity is visible on mobile.
+
+describe("MobileChat — Agent Comms render parity (#231)", () => {
+  beforeEach(() => {
+    mockStoreState.nodes = [onlineNode];
+  });
+
+  it("mounts AgentCommsPanel on the Agent Comms tab (not the old placeholder)", async () => {
+    const getSpy = vi.spyOn(api, "get");
+    // 1st GET: useChatHistory (My Chat) on mount.
+    getSpy.mockResolvedValueOnce({ messages: [], reached_end: true });
+    // 2nd GET: AgentCommsPanel's activity load when the tab is shown.
+    // Empty list → panel renders its own empty state, which still
+    // proves AgentCommsPanel mounted (vs. the removed placeholder).
+    getSpy.mockResolvedValueOnce([]);
+
+    let rr: ReturnType<typeof renderChat>;
+    await act(async () => {
+      rr = renderChat(mockAgentId);
+    });
+    const { container } = rr!;
+
+    const commsTab = Array.from(container.querySelectorAll("button")).find(
+      (b) => b.textContent?.trim() === "Agent Comms",
+    );
+    expect(commsTab).toBeTruthy();
+    await act(async () => {
+      commsTab!.click();
+    });
+
+    await waitFor(() => {
+      const text = container.textContent ?? "";
+      // The panel's empty state — proves AgentCommsPanel mounted.
+      expect(text).toContain("No agent-to-agent communications yet.");
+    });
+    // The old hard-coded placeholder must be gone.
+    expect(container.textContent ?? "").not.toContain(
+      "peer-to-peer A2A traffic surfaces in the Comms tab",
+    );
+    // The panel hit its activity endpoint.
+    expect(getSpy).toHaveBeenCalledWith(
+      expect.stringContaining(`/workspaces/${mockAgentId}/activity`),
+    );
+  });
+
+  it("renders a peer message on the Agent Comms tab", async () => {
+    const getSpy = vi.spyOn(api, "get");
+    getSpy.mockResolvedValueOnce({ messages: [], reached_end: true });
+    // a2a_receive from a peer → AgentCommsPanel.toCommMessage maps it
+    // to an inbound bubble with the request text.
+    getSpy.mockResolvedValueOnce([
+      {
+        id: "act-1",
+        activity_type: "a2a_receive",
+        source_id: "peer-ws-uuid",
+        target_id: mockAgentId,
+        method: "message/send",
+        summary: "peer asked something",
+        request_body: { task: "Please review PR 42" },
+        response_body: null,
+        status: "ok",
+        created_at: new Date().toISOString(),
+      },
+    ]);
+
+    let rr: ReturnType<typeof renderChat>;
+    await act(async () => {
+      rr = renderChat(mockAgentId);
+    });
+    const { container } = rr!;
+
+    const commsTab = Array.from(container.querySelectorAll("button")).find(
+      (b) => b.textContent?.trim() === "Agent Comms",
+    );
+    await act(async () => {
+      commsTab!.click();
+    });
+
+    await waitFor(() => {
+      expect(container.textContent ?? "").toContain("Please review PR 42");
+    });
+  });
+});
@@ -93,6 +93,24 @@ describe("MobileSpawn — render", () => {
    expect(input).toBeTruthy();
  });

+  // Regression #224 / #225: the agent-name input must render with a
+  // font-size ≥ 16px. iOS Safari and PWAs auto-zoom the viewport when a
+  // focused input has a computed font-size below 16px — the layout
+  // jumps and the page looks broken until the user pinches back.
+  it("renders the name input at font-size 16px or greater (iOS focus-zoom regression)", () => {
+    apiGetSpy.mockResolvedValue(mockTemplates);
+    render(<MobileSpawn dark={true} onClose={vi.fn()} />);
+    const input = document.querySelector(
+      'input[aria-label="Agent name"]',
+    ) as HTMLInputElement | null;
+    expect(input).toBeTruthy();
+    // Parse the inline style font-size — jsdom doesn't run a layout
+    // engine, so getComputedStyle reports the inline value verbatim.
+    const fs = Number.parseFloat(input!.style.fontSize);
+    expect(Number.isFinite(fs)).toBe(true);
+    expect(fs).toBeGreaterThanOrEqual(16);
+  });
+
  it("renders all 4 tier buttons", () => {
    apiGetSpy.mockResolvedValue(mockTemplates);
    render(<MobileSpawn dark={true} onClose={vi.fn()} />);
@@ -133,6 +133,7 @@ export function TabBar({
            aria-label={t.label}
            onClick={() => onChange(t.id)}
            onKeyDown={(e) => handleKeyDown(e, idx)}
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              background: "none",
              border: "none",
@@ -291,6 +292,7 @@ export function AgentCard({
      data-testid="workspace-card"
      aria-label={`${agent.name}, status: ${agent.status}, tier ${agent.tier}${agent.remote ? ", remote" : ""}`}
      onClick={onClick}
+      className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
      style={{
        display: "block",
        width: "100%",
@@ -444,6 +446,7 @@ export function FilterChips({
            type="button"
            aria-checked={on}
            onClick={() => onChange(o.id)}
+            className="focus:outline-none focus-visible:ring-2 focus-visible:ring-emerald-500 focus-visible:ring-offset-2 focus-visible:ring-offset-zinc-100 dark:focus-visible:ring-offset-zinc-900"
            style={{
              display: "inline-flex",
              alignItems: "center",
@@ -0,0 +1,181 @@
+'use client';
+
+import { useCallback, useEffect, useState } from 'react';
+import { api } from '@/lib/api';
+import { fetchSession, type Session } from '@/lib/auth';
+import { getTenantSlug } from '@/lib/tenant';
+import { Spinner } from '@/components/Spinner';
+
+/**
+ * Organization-identity surface inside SettingsPanel.
+ *
+ * Closes a chronic UX gap where users (and our own AI agents) had to
+ * call /cp/auth/me or /cp/orgs from browser devtools to read their
+ * org_id UUID. Now: a copy-buttoned view of name + slug + UUID for the
+ * currently-active org, plus a switcher list when the user belongs to
+ * multiple orgs.
+ *
+ * Data path:
+ *   1. fetchSession() → /cp/auth/me → current org_id
+ *   2. api.get('/cp/orgs') → list of all orgs the user belongs to
+ *   3. Match by id === session.org_id; fall back to host-slug match
+ *      if the session probe loses the race.
+ *
+ * Read-only — this tab never mutates. Org creation/switching lives at
+ * /orgs (the post-signup landing page).
+ */
+
+interface Org {
+  id: string;
+  slug: string;
+  name: string;
+  status?: string;
+}
+
+// /cp/orgs may return a bare array or {orgs: []} — see orgs/page.tsx
+// for the same defensive unwrap.
+type OrgsResponse = Org[] | { orgs?: Org[] };
+
+export function OrgInfoTab() {
+  const [orgs, setOrgs] = useState<Org[] | null>(null);
+  const [session, setSession] = useState<Session | null>(null);
+  const [error, setError] = useState<string | null>(null);
+  const [loading, setLoading] = useState(true);
+
+  useEffect(() => {
+    let cancelled = false;
+    (async () => {
+      try {
+        const [sess, body] = await Promise.all([
+          fetchSession().catch(() => null),
+          api.get<OrgsResponse>('/cp/orgs'),
+        ]);
+        if (cancelled) return;
+        setSession(sess);
+        setOrgs(Array.isArray(body) ? body : body.orgs ?? []);
+      } catch (e) {
+        if (!cancelled) setError(e instanceof Error ? e.message : 'Failed to load org info');
+      } finally {
+        if (!cancelled) setLoading(false);
+      }
+    })();
+    return () => {
+      cancelled = true;
+    };
+  }, []);
+
+  const tenantSlug = getTenantSlug();
+  const currentOrg =
+    orgs?.find((o) => session && o.id === session.org_id) ??
+    orgs?.find((o) => tenantSlug && o.slug === tenantSlug) ??
+    null;
+  const otherOrgs = orgs?.filter((o) => o.id !== currentOrg?.id) ?? [];
+
+  if (loading) {
+    return (
+      <div
+        role="status"
+        aria-live="polite"
+        className="flex items-center justify-center gap-2 py-6 text-ink-mid text-xs"
+      >
+        <Spinner /> Loading organization…
+      </div>
+    );
+  }
+  if (error) {
+    return (
+      <div className="p-4">
+        <div className="px-3 py-2 bg-red-950/40 border border-red-800/50 rounded-lg text-[10px] text-bad">
+          {error}
+        </div>
+      </div>
+    );
+  }
+  if (!currentOrg) {
+    return (
+      <div className="p-4">
+        <p className="text-xs text-ink-mid">
+          No organization found for this session. If this is unexpected, sign out and back in, or visit{' '}
+          <a href="/orgs" className="underline">/orgs</a>.
+        </p>
+      </div>
+    );
+  }
+
+  return (
+    <div className="p-4 space-y-4">
+      <div>
+        <h3 className="text-sm font-semibold text-ink mb-1">Current Organization</h3>
+        <p className="text-[10px] text-ink-mid leading-relaxed">
+          IDs you can paste into API calls, support tickets, or CLI arguments. The UUID never changes;
+          the slug is the URL subdomain.
+        </p>
+      </div>
+      <OrgIdentityCard org={currentOrg} highlighted />
+      {otherOrgs.length > 0 && (
+        <div className="space-y-2 pt-2">
+          <h4 className="text-[11px] font-semibold text-ink-mid uppercase tracking-wider">
+            Your other organizations ({otherOrgs.length})
+          </h4>
+          {otherOrgs.map((o) => (
+            <OrgIdentityCard key={o.id} org={o} />
+          ))}
+        </div>
+      )}
+    </div>
+  );
+}
+
+function OrgIdentityCard({ org, highlighted }: { org: Org; highlighted?: boolean }) {
+  return (
+    <div
+      className={`rounded-lg border p-3 space-y-2 ${
+        highlighted ? 'border-accent/40 bg-accent-strong/5' : 'border-line/40 bg-surface-card/40'
+      }`}
+      data-testid={`org-card-${org.slug}`}
+    >
+      <div className="flex items-baseline justify-between gap-2">
+        <span className="text-[12px] font-medium text-ink truncate">{org.name}</span>
+        {org.status && (
+          <span className="text-[9px] text-ink-mid uppercase tracking-wider shrink-0">{org.status}</span>
+        )}
+      </div>
+      <IdentityRow label="Slug" value={org.slug} />
+      <IdentityRow label="UUID" value={org.id} mono />
+    </div>
+  );
+}
+
+function IdentityRow({ label, value, mono }: { label: string; value: string; mono?: boolean }) {
+  const [copied, setCopied] = useState(false);
+  const onCopy = useCallback(() => {
+    // Best-effort: jsdom + old Safari throw synchronously on writeText.
+    try {
+      navigator.clipboard.writeText(value);
+    } catch {
+      /* user can still triple-click select */
+    }
+    setCopied(true);
+    setTimeout(() => setCopied(false), 2000);
+  }, [value]);
+  return (
+    <div className="flex items-center gap-2">
+      <span className="text-[10px] text-ink-mid w-10 shrink-0">{label}</span>
+      <code
+        className={`flex-1 text-[11px] text-ink bg-surface-sunken/60 px-2 py-1 rounded select-all break-all ${
+          mono ? 'font-mono' : ''
+        }`}
+      >
+        {value}
+      </code>
+      <button
+        type="button"
+        onClick={onCopy}
+        aria-label={`Copy ${label}`}
+        className="shrink-0 px-2 py-1 bg-surface-card/60 hover:bg-surface-card border border-line/40 rounded text-[10px] text-ink-mid hover:text-ink transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
+      >
+        {copied ? 'Copied' : 'Copy'}
+      </button>
+    </div>
+  );
+}
@@ -160,14 +160,14 @@ export function OrgTokensTab() {
            </code>
            <button
              onClick={handleCopy}
-              className="shrink-0 px-2 py-1.5 bg-emerald-800/40 hover:bg-emerald-700/50 border border-emerald-700/40 rounded text-[10px] text-good transition-colors"
+              className="shrink-0 px-2 py-1.5 bg-emerald-800/40 hover:bg-emerald-700/50 border border-emerald-700/40 rounded text-[10px] text-good transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
            >
              {copied ? 'Copied' : 'Copy'}
            </button>
          </div>
          <button
            onClick={() => setNewToken(null)}
-            className="text-[9px] text-good/60 hover:text-good transition-colors"
+            className="text-[9px] text-good/60 hover:text-good transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
          >
            Dismiss
          </button>
@@ -219,7 +219,7 @@ export function OrgTokensTab() {
              </div>
              <button
                onClick={() => setRevokeTarget(t)}
-                className="text-[10px] text-bad/70 hover:text-bad transition-colors px-2 py-1 shrink-0"
+                className="text-[10px] text-bad/70 hover:text-bad transition-colors px-2 py-1 shrink-0 focus:outline-none focus-visible:ring-2 focus-visible:ring-red-400 focus-visible:ring-offset-1"
              >
                Revoke
              </button>
@@ -3,16 +3,24 @@ import { useState, useCallback, useRef, useEffect } from 'react';
 import type { Secret, SecretGroup } from '@/types/secrets';
 import { useSecretsStore } from '@/stores/secrets-store';
 import { StatusBadge } from '@/components/ui/StatusBadge';
-import { RevealToggle } from '@/components/ui/RevealToggle';
 import { KeyValueField } from '@/components/ui/KeyValueField';
 import { ValidationHint } from '@/components/ui/ValidationHint';
 import { TestConnectionButton } from '@/components/ui/TestConnectionButton';
 import { validateSecretValue } from '@/lib/validation/secret-formats';
 import { SERVICES } from '@/lib/services';

-const AUTO_HIDE_MS = 30_000;
 const VALIDATION_DEBOUNCE_MS = 400;

+// Secret values are write-only from the browser: the server List endpoint
+// "Never exposes values", there is no per-secret decrypt route, and the
+// only decrypted path (GET /secrets/values) is bulk + token-gated for
+// remote agents. The old eye/RevealToggle was a dead affordance — it
+// flipped its own icon but could never reveal anything, which read as
+// "this doesn't work" (esp. once clicked → eye-with-slash). We show an
+// honest static indicator instead; rotation is via Edit.
+const WRITE_ONLY_TITLE =
+  'Value is write-only and cannot be revealed — use Edit to replace/rotate it';
+
 interface SecretRowProps {
  secret: Secret;
  workspaceId: string;
@@ -31,28 +39,12 @@ export function SecretRow({ secret, workspaceId }: SecretRowProps) {
  const setSecretStatus = useSecretsStore((s) => s.setSecretStatus);

  const isEditing = editingKey === secret.name;
-  const [revealed, setRevealed] = useState(false);
  const [editValue, setEditValue] = useState('');
  const [validationError, setValidationError] = useState<string | null>(null);
  const [isSaving, setIsSaving] = useState(false);
  const [saveError, setSaveError] = useState<string | null>(null);
  const debounceRef = useRef<ReturnType<typeof setTimeout>>(undefined);
  const editBtnRef = useRef<HTMLButtonElement>(null);
-  const revealTimerRef = useRef<ReturnType<typeof setTimeout>>(undefined);
-
-  // Auto-hide revealed value after 30s
-  useEffect(() => {
-    if (revealed) {
-      clearTimeout(revealTimerRef.current);
-      revealTimerRef.current = setTimeout(() => setRevealed(false), AUTO_HIDE_MS);
-      return () => clearTimeout(revealTimerRef.current);
-    }
-  }, [revealed]);
-
-  // Reset revealed state when panel closes (session-only)
-  useEffect(() => {
-    return () => setRevealed(false);
-  }, []);

  // Debounced validation
  useEffect(() => {
@@ -133,11 +125,15 @@ export function SecretRow({ secret, workspaceId }: SecretRowProps) {
          {secret.masked_value}
        </span>
        <div className="secret-row__actions">
-          <RevealToggle
-            revealed={revealed}
-            onToggle={() => setRevealed((r) => !r)}
-            label={`Toggle reveal ${secret.name}`}
-          />
+          <span
+            data-testid="write-only-indicator"
+            className="secret-row__write-only"
+            role="img"
+            aria-label={`${secret.name} value is write-only and cannot be revealed; use Edit to replace it`}
+            title={WRITE_ONLY_TITLE}
+          >
+            🔒
+          </span>
          <StatusBadge status={secret.status} />
          <button
            type="button"
--- a/Show More
+++ b/Show More