fix(a2a-mcp): use readline() not read(65536) for pipe-safe stdio

a2a_mcp_server.py main()'s stdio read loop used
`await loop.run_in_executor(None, stdin.read, 65536)`. On a PIPE,
read(n) blocks until n bytes accumulate OR EOF. A live MCP client
(openclaw bundle-mcp, Claude Code, Cursor) sends one ~150-byte
newline-delimited request and keeps stdin OPEN waiting for the reply,
so neither condition is met: the server never parses `initialize` and
the client times out (~30s; openclaw: "MCP error -32000: Connection
closed"). This silently broke peer visibility for every pipe-spawned
MCP host while passing all existing stdio tests, which only fed stdin
from a regular file or a heredoc-pipe that CLOSES (EOF returns
immediately). readline() returns as soon as one newline-delimited
line is available — exactly the JSON-RPC framing — and is
backward-compatible with the EOF/file cases.

Root cause of the 2026-05-15 openclaw peer-visibility outage
(workspace 95744c11): the molecule MCP server could not complete the
handshake over openclaw's stdio pipe, so the agent fell back to
native sessions_list. The openclaw template adapter fix
(template-openclaw#16) works around this via HTTP transport; this
patch fixes the stdio root cause so stdio works for all CLI MCP hosts.

Regression coverage:
- tests/test_a2a_mcp_server.py::TestStdioKeepOpenPipe — spawns the
  real a2a_mcp_server.py, writes one request over a pipe, and
  DELIBERATELY keeps stdin open. FAILS (15s timeout, empty response)
  on read(65536); PASSES on readline(). Verified both directions.
- ci-mcp-stdio-transport.yml: new "pipe held OPEN, no EOF" step that
  reproduces the literal openclaw failure (the prior steps only
  exercised EOF-closing stdin, which is why the outage shipped green).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

.gitea/ci-refire

@@ -0,0 +1 @@
+refire:1778784369

.gitea/scripts/ci-required-drift.py

@@ -203,12 +203,17 @@ def ci_jobs_all(ci_doc: dict) -> set[str]:
 
 def ci_job_names(ci_doc: dict) -> set[str]:
     """Set of job keys in ci.yml MINUS the sentinel itself MINUS jobs
-    whose `if:` gates on `github.event_name` (those are event-scoped
-    and can legitimately be `skipped` for a given trigger; if we
-    required them under the sentinel `needs:`, every PR-only job
+    whose `if:` gates on `github.event_name` or `github.ref` (those are
+    event-scoped and can legitimately be `skipped` for a given trigger;
+    if we required them under the sentinel `needs:`, every PR-only job
     would be `skipped` on push and the sentinel would interpret
     `skipped != success` as failure). RFC §4 spec.
 
+    `github.ref` is the companion gate for jobs that run only on direct
+    pushes to specific branches (e.g. `github.ref == 'refs/heads/main'`).
+    These never execute in a PR context, so flagging them as missing
+    from `all-required.needs:` is a false positive (mc#958 / mc#959).
+
     Used for F1 (jobs missing from sentinel needs). NOT used for F1b
     (typos in needs) — see `ci_jobs_all` for that."""
     jobs = ci_doc.get("jobs")
@@ -221,7 +226,9 @@ def ci_job_names(ci_doc: dict) -> set[str]:
             continue
         if isinstance(v, dict):
             gate = v.get("if")
-            if isinstance(gate, str) and "github.event_name" in gate:
+            if isinstance(gate, str) and (
+                "github.event_name" in gate or "github.ref" in gate
+            ):
                 continue
         names.add(k)
     return names

.gitea/scripts/gitea-merge-queue.py

@@ -47,6 +47,15 @@ REQUIRED_CONTEXTS_RAW = _env(
         "sop-checklist / all-items-acked (pull_request)"
     ),
 )
+# Required contexts for push (main/staging) runs. The push CI uses the same
+# aggregator names with " (push)" suffix. Checking these explicitly instead of
+# the combined state avoids false-pause when non-blocking jobs (e.g. Platform
+# Go with continue-on-error: true due to mc#774) have failed — their failures
+# pollute the combined state but do not block merges.
+PUSH_REQUIRED_CONTEXTS_RAW = _env(
+    "PUSH_REQUIRED_CONTEXTS",
+    default="CI / all-required (push)",
+)
 
 OWNER, NAME = (REPO.split("/", 1) + [""])[:2] if REPO else ("", "")
 API = f"https://{GITEA_HOST}/api/v1" if GITEA_HOST else ""
@@ -118,16 +127,24 @@ def required_contexts(raw: str) -> list[str]:
     return [part.strip() for part in raw.split(",") if part.strip()]
 
 
+def push_required_contexts() -> list[str]:
+    """Required contexts for push (branch) CI runs. See PUSH_REQUIRED_CONTEXTS_RAW."""
+    return required_contexts(PUSH_REQUIRED_CONTEXTS_RAW)
+
+
 def status_state(status: dict) -> str:
     return str(status.get("status") or status.get("state") or "").lower()
 
 
 def latest_statuses_by_context(statuses: list[dict]) -> dict[str, dict]:
+    # Gitea /statuses endpoint returns entries in ascending id order (oldest
+    # first). We need the LAST occurrence of each context, so iterate in
+    # reverse to prefer newer entries.
     latest: dict[str, dict] = {}
-    for status in statuses:
+    for status in reversed(statuses):
         context = status.get("context")
-        if isinstance(context, str) and context not in latest:
-            latest[context] = status
+        if isinstance(context, str):
+            latest[context] = status  # overwrite: reverse order → newest wins
     return latest
 
 
@@ -193,16 +210,23 @@ def evaluate_merge_readiness(
     required_contexts: list[str],
     pr_has_current_base: bool,
 ) -> MergeDecision:
-    main_state = str(main_status.get("state") or "").lower()
-    if main_state != "success":
-        return MergeDecision(False, "pause", f"main status is {main_state or 'missing'}")
+    # Check push-required contexts explicitly instead of combined state.
+    # Combined state can be "failure" due to non-blocking jobs
+    # (continue-on-error: true) that don't actually gate merges.
+    # CI / all-required (push) is the authoritative gate — it respects
+    # continue-on-error and correctly aggregates all blocking failures.
+    main_latest = latest_statuses_by_context(main_status.get("statuses") or [])
+    main_ok, main_bad = required_contexts_green(main_latest, push_required_contexts())
+    if not main_ok:
+        return MergeDecision(False, "pause", "main required contexts not green: " + ", ".join(main_bad))
     if not pr_has_current_base:
         return MergeDecision(False, "update", "PR head does not contain current main")
 
-    pr_state = str(pr_status.get("state") or "").lower()
-    if pr_state != "success":
-        return MergeDecision(False, "wait", f"PR combined status is {pr_state or 'missing'}")
-
+    # Check explicit required contexts instead of combined state. Combined state
+    # can be "failure" due to non-blocking jobs with continue-on-error: true
+    # (e.g. publish-runtime-autobump/pr-validate, qa-review on stale tokens).
+    # 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)
     if not ok:
@@ -220,10 +244,37 @@ def get_branch_head(branch: str) -> str:
 
 
 def get_combined_status(sha: str) -> dict:
-    _, body = api("GET", f"/repos/{OWNER}/{NAME}/commits/{sha}/status")
-    if not isinstance(body, dict):
+    """Combined status + all individual statuses for `sha`.
+
+    The /status endpoint caps the `statuses` array at 30 entries (Gitea
+    default page size), so we fetch the full list via /statuses with a
+    higher limit. The combined `state` still comes from /status.
+    """
+    _, combined = api("GET", f"/repos/{OWNER}/{NAME}/commits/{sha}/status")
+    if not isinstance(combined, dict):
         raise ApiError(f"status for {sha} response not object")
-    return body
+    # 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).
+    try:
+        _, all_statuses = api(
+            "GET",
+            f"/repos/{OWNER}/{NAME}/commits/{sha}/statuses",
+            query={"limit": "50"},
+        )
+        if isinstance(all_statuses, list):
+            combined["statuses"] = 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
+    return combined
 
 
 def list_queued_issues() -> list[dict]:
@@ -294,8 +345,12 @@ def process_once(*, dry_run: bool = False) -> int:
     contexts = required_contexts(REQUIRED_CONTEXTS_RAW)
     main_sha = get_branch_head(WATCH_BRANCH)
     main_status = get_combined_status(main_sha)
-    if str(main_status.get("state") or "").lower() != "success":
-        print(f"::notice::queue paused: {WATCH_BRANCH}@{main_sha[:8]} is not green")
+    # Check push-required contexts explicitly instead of combined state.
+    # See evaluate_merge_readiness for rationale.
+    main_latest = latest_statuses_by_context(main_status.get("statuses") or [])
+    main_ok, main_bad = required_contexts_green(main_latest, push_required_contexts())
+    if not main_ok:
+        print(f"::notice::queue paused: {WATCH_BRANCH}@{main_sha[:8]} required contexts not green: {', '.join(main_bad)}")
         return 0
 
     issue = choose_next_queued_issue(
@@ -362,7 +417,21 @@ def main() -> int:
     parser.add_argument("--dry-run", action="store_true")
     args = parser.parse_args()
     _require_runtime_env()
-    return process_once(dry_run=args.dry_run)
+    try:
+        return process_once(dry_run=args.dry_run)
+    except ApiError as exc:
+        # API errors (401/403/404/500) are transient for a queue tick —
+        # log and exit 0 so the workflow is not marked failed and the next
+        # tick can retry. Returning non-zero would permanently fail the
+        # workflow run, blocking future ticks.
+        sys.stderr.write(f"::error::queue API error: {exc}\n")
+        return 0
+    except urllib.error.URLError as exc:
+        sys.stderr.write(f"::error::queue network error: {exc}\n")
+        return 0
+    except TimeoutError as exc:
+        sys.stderr.write(f"::error::queue timeout: {exc}\n")
+        return 0
 
 
 if __name__ == "__main__":

.gitea/scripts/lint-workflow-yaml.py

@@ -36,6 +36,9 @@ Rules (4 fatal + 1 fatal cross-file + 1 heuristic-warn):
      raw `.error` fields into CI logs/summaries.
   9. Production deploy/redeploy workflows must expose an operational control:
      kill switch for auto deploys or rollback tag for manual deploys.
+  10. Docker health checks must not run `docker info | head` under pipefail.
+      `head` closes the pipe early, `docker info` can exit nonzero from
+      SIGPIPE, and the step can falsely report Docker daemon failure.
 
 Per `feedback_smoke_test_vendor_truth_not_shape_match`: fixtures used to
 validate this lint must mirror real Gitea 1.22.6 YAML semantics, not
@@ -225,6 +228,24 @@ def _iter_uses(doc: Any) -> Iterable[str]:
                 yield step["uses"]
 
 
+def _iter_run_blocks(doc: Any) -> Iterable[str]:
+    """Yield every shell `run:` block from job steps in a workflow document."""
+    if not isinstance(doc, dict):
+        return
+    jobs = doc.get("jobs")
+    if not isinstance(jobs, dict):
+        return
+    for job in jobs.values():
+        if not isinstance(job, dict):
+            continue
+        steps = job.get("steps")
+        if not isinstance(steps, list):
+            continue
+        for step in steps:
+            if isinstance(step, dict) and isinstance(step.get("run"), str):
+                yield step["run"]
+
+
 def check_cross_repo_uses(filename: str, doc: Any) -> list[str]:
     """Return per-violation error lines for cross-repo `uses:` references."""
     errors: list[str] = []
@@ -264,6 +285,10 @@ GITHUB_API_REF_RE = re.compile(
 
 PROD_CP_URL_RE = re.compile(r"https://api\.moleculesai\.app\b")
 REDEPLOY_FLEET_RE = re.compile(r"\b/cp/admin/tenants/redeploy-fleet\b")
+RUN_SETS_PIPEFAIL_RE = re.compile(r"(?m)^\s*set\s+-[^\n]*o\s+pipefail\b")
+DOCKER_INFO_HEAD_PIPE_RE = re.compile(
+    r"(?m)^\s*docker\s+info\b[^\n|]*\|\s*head\b"
+)
 RAW_CP_RESPONSE_RE = re.compile(
     r"""(?x)
     (?:\bjq\s+\.\s+["']?\$HTTP_RESPONSE["']?)
@@ -383,6 +408,30 @@ def check_production_operational_control(filename: str, raw: str) -> list[str]:
     return errors
 
 
+# ---------------------------------------------------------------------------
+# Rule 10 — docker info piped to head under pipefail
+# ---------------------------------------------------------------------------
+
+def check_docker_info_head_pipefail(filename: str, doc: Any) -> list[str]:
+    errors: list[str] = []
+    for run_block in _iter_run_blocks(doc):
+        if not (
+            RUN_SETS_PIPEFAIL_RE.search(run_block)
+            and DOCKER_INFO_HEAD_PIPE_RE.search(run_block)
+        ):
+            continue
+        errors.append(
+            f"::error file={filename}::Rule 10 (FATAL): workflow runs "
+            f"`docker info | head` after enabling `pipefail`. `head` can "
+            f"close the pipe early, making `docker info` exit nonzero and "
+            f"falsely fail the Docker daemon health check. Capture "
+            f"`docker_info=\"$(docker info 2>&1)\"` first, then print a "
+            f"bounded preview with `printf ... | sed -n '1,5p'`."
+        )
+        break
+    return errors
+
+
 # ---------------------------------------------------------------------------
 # Driver
 # ---------------------------------------------------------------------------
@@ -436,6 +485,7 @@ def main(argv: list[str] | None = None) -> int:
         fatal_errors.extend(check_production_concurrency(rel, doc, raw))
         fatal_errors.extend(check_production_raw_response_logging(rel, raw))
         fatal_errors.extend(check_production_operational_control(rel, raw))
+        fatal_errors.extend(check_docker_info_head_pipefail(rel, doc))
         warnings.extend(check_github_server_url_missing(rel, doc, raw))
 
     # Cross-file checks

.gitea/scripts/review-check.sh

@@ -101,9 +101,10 @@ printf 'header = "Authorization: token %s"\n' "$GITEA_TOKEN" > "$CURL_AUTH_FILE"
 PR_JSON=$(mktemp)
 REVIEWS_JSON=$(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"
+  rm -f "$CURL_AUTH_FILE" "$PR_JSON" "$REVIEWS_JSON" "$TEAM_PROBE_TMP" "${NA_STATUSES_TMP-}"
 }
 trap cleanup EXIT
 
@@ -143,6 +144,42 @@ if [ -z "$PR_AUTHOR" ] || [ -z "$PR_HEAD_SHA" ]; then
   exit 1
 fi
 
+# --- RFC#324 §N/A follow-up: check N/A declarations status ---
+# sop-checklist.py posts `sop-checklist / na-declarations (pull_request)`
+# status when a peer posts /sop-n/a <gate>. If our gate is declared N/A,
+# the requirement for a Gitea APPROVE review is waived.
+NA_STATUSES_TMP=$(mktemp)
+HTTP_CODE=$(curl -sS -o "$NA_STATUSES_TMP" -w '%{http_code}' \
+  -K "$CURL_AUTH_FILE" "${API}/repos/${OWNER}/${NAME}/statuses/${PR_HEAD_SHA}")
+debug "statuses/${PR_HEAD_SHA} → HTTP ${HTTP_CODE}"
+
+if [ "$HTTP_CODE" = "200" ]; then
+  # Gitea returns statuses as array; look for the na-declarations context.
+  # jq: find all statuses where context == "sop-checklist / na-declarations (pull_request)"
+  # and state == "success". Extract the description field.
+  NA_DESC=$(jq -r '
+    .[] |
+    select(.context == "sop-checklist / na-declarations (pull_request)") |
+    select(.state == "success") |
+    .description
+  ' "$NA_STATUSES_TMP" 2>/dev/null | head -1)
+
+  if [ -n "$NA_DESC" ] && [ "$NA_DESC" != "null" ]; then
+    debug "na-declarations status found: ${NA_DESC}"
+    # Check if our gate appears in the N/A description.
+    # The description format is "N/A: qa-review, security-review" or similar.
+    if echo "$NA_DESC" | grep -iq "\\b${TEAM}-review\\b"; then
+      echo "::notice::${TEAM}-review N/A — gate declared not-applicable via /sop-n/a: ${NA_DESC}"
+      echo "::notice::PR ${PR_NUMBER} passes ${TEAM}-review via N/A declaration"
+      rm -f "$NA_STATUSES_TMP"
+      exit 0
+    fi
+  fi
+else
+  debug "could not fetch statuses (HTTP ${HTTP_CODE}) — proceeding with normal eval"
+fi
+rm -f "$NA_STATUSES_TMP"
+
 # --- Fetch all reviews on the PR ---
 HTTP_CODE=$(curl -sS -o "$REVIEWS_JSON" -w '%{http_code}' \
   -K "$CURL_AUTH_FILE" "${API}/repos/${OWNER}/${NAME}/pulls/${PR_NUMBER}/reviews")

.gitea/scripts/review-refire-status.sh

@@ -0,0 +1,81 @@
+#!/usr/bin/env bash
+# Re-run review-check.sh for a slash-command refire and post the protected
+# pull_request status context to the PR head SHA.
+
+set -euo pipefail
+
+: "${GITEA_TOKEN:?GITEA_TOKEN required}"
+: "${GITEA_HOST:?GITEA_HOST required}"
+: "${REPO:?REPO required}"
+: "${PR_NUMBER:?PR_NUMBER required}"
+: "${TEAM:?TEAM required}"
+
+OWNER="${REPO%%/*}"
+NAME="${REPO##*/}"
+API="https://${GITEA_HOST}/api/v1"
+CONTEXT="${TEAM}-review / approved (pull_request)"
+TARGET_URL="https://${GITEA_HOST}/${OWNER}/${NAME}/pulls/${PR_NUMBER}"
+
+authfile=$(mktemp)
+prfile=$(mktemp)
+postfile=$(mktemp)
+# shellcheck disable=SC2329 # invoked by EXIT trap
+cleanup() {
+  rm -f "$authfile" "$prfile" "$postfile"
+}
+trap cleanup EXIT
+
+chmod 600 "$authfile"
+printf 'header = "Authorization: token %s"\n' "$GITEA_TOKEN" > "$authfile"
+
+code=$(curl -sS -o "$prfile" -w '%{http_code}' -K "$authfile" \
+  "${API}/repos/${OWNER}/${NAME}/pulls/${PR_NUMBER}")
+if [ "$code" != "200" ]; then
+  echo "::error::GET /pulls/${PR_NUMBER} returned HTTP ${code}"
+  head -c 200 "$prfile" >&2 || true
+  exit 1
+fi
+
+head_sha=$(jq -r '.head.sha // ""' "$prfile")
+state=$(jq -r '.state // ""' "$prfile")
+if [ -z "$head_sha" ] || [ "$head_sha" = "null" ]; then
+  echo "::error::Could not resolve PR head SHA for PR ${PR_NUMBER}"
+  exit 1
+fi
+if [ "$state" != "open" ]; then
+  echo "::notice::PR ${PR_NUMBER} is ${state}; ${TEAM}-review refire is a no-op"
+  exit 0
+fi
+
+set +e
+bash .gitea/scripts/review-check.sh
+rc=$?
+set -e
+
+if [ "$rc" -eq 0 ]; then
+  status_state="success"
+  description="Refired via /${TEAM}-recheck by ${COMMENT_AUTHOR:-unknown}"
+else
+  status_state="failure"
+  description="Refired via /${TEAM}-recheck; ${TEAM}-review failed"
+fi
+
+body=$(jq -nc \
+  --arg state "$status_state" \
+  --arg context "$CONTEXT" \
+  --arg description "$description" \
+  --arg target_url "$TARGET_URL" \
+  '{state:$state, context:$context, description:$description, target_url:$target_url}')
+
+code=$(curl -sS -o "$postfile" -w '%{http_code}' -X POST \
+  -K "$authfile" -H "Content-Type: application/json" \
+  -d "$body" \
+  "${API}/repos/${OWNER}/${NAME}/statuses/${head_sha}")
+if [ "$code" != "200" ] && [ "$code" != "201" ]; then
+  echo "::error::POST /statuses/${head_sha} returned HTTP ${code}"
+  head -c 200 "$postfile" >&2 || true
+  exit 1
+fi
+
+echo "::notice::posted ${status_state} for context=\"${CONTEXT}\" on sha=${head_sha}"
+exit "$rc"

.gitea/scripts/sop-checklist.py

@@ -1,11 +1,11 @@
 #!/usr/bin/env python3
-# sop-checklist-gate — evaluate whether a PR has peer-acked each
+# sop-checklist — evaluate whether a PR has peer-acked each
 # SOP-checklist item. Posts a commit-status that branch protection
 # can require.
 #
 # RFC#351 Step 2 of 6 (implementation MVP).
 #
-# Invoked by .gitea/workflows/sop-checklist-gate.yml on:
+# Invoked by .gitea/workflows/sop-checklist.yml on:
 #   - pull_request_target: [opened, edited, synchronize, reopened]
 #   - issue_comment:       [created, edited, deleted]
 #
@@ -118,17 +118,19 @@ _DIRECTIVE_RE = re.compile(
 def parse_directives(
     comment_body: str,
     numeric_aliases: dict[int, str],
-) -> list[tuple[str, str, str]]:
+) -> tuple[list[tuple[str, str, str]], list]:
     """Extract /sop-ack and /sop-revoke directives from a comment body.
 
-    Returns a list of (kind, canonical_slug, note) tuples where:
-      kind is "sop-ack" or "sop-revoke"
-      canonical_slug is the normalized form (or "" if unparseable)
-      note is the trailing free-text (may be "")
+    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)
     """
     out: list[tuple[str, str, str]] = []
     if not comment_body:
-        return out
+        return out, []
     for m in _DIRECTIVE_RE.finditer(comment_body):
         kind = m.group(1)
         raw_slug = (m.group(2) or "").strip()
@@ -159,7 +161,7 @@ def parse_directives(
         # 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
+    return out, []
 
 
 # ---------------------------------------------------------------------------
@@ -249,7 +251,8 @@ def compute_ack_state(
         user = (c.get("user") or {}).get("login", "")
         if not user:
             continue
-        for kind, slug, _note in parse_directives(body, numeric_aliases):
+        directives, _na = parse_directives(body, numeric_aliases)
+        for kind, slug, _note in directives:
             if not slug:
                 unparseable_per_user[user] = unparseable_per_user.get(user, 0) + 1
                 continue

.gitea/scripts/status-reaper.py

@@ -133,6 +133,9 @@ PUSH_COMPENSATION_DESCRIPTION = (
     "Compensated by status-reaper (workflow has no push: trigger; "
     "Gitea 1.22.6 hardcoded-suffix bug — see .gitea/scripts/status-reaper.py)"
 )
+# Backward-compatible alias for older tests/tooling that predate the split
+# between push-suffix compensation and pull-request-shadow compensation.
+COMPENSATION_DESCRIPTION = PUSH_COMPENSATION_DESCRIPTION
 PR_SHADOW_COMPENSATION_DESCRIPTION = (
     "Compensated by status-reaper (default-branch pull_request status "
     "shadowed by successful push status on same SHA; see "
@@ -611,11 +614,10 @@ def list_recent_commit_shas(branch: str, limit: int) -> list[str]:
     (verified via vendor-truth probe 2026-05-11 against
     git.moleculesai.app — `feedback_smoke_test_vendor_truth_not_shape_match`).
 
-    Raises ApiError on non-2xx OR on unexpected response shape. This is
-    a HARD halt — without the commit list the sweep can't proceed. (The
-    per-SHA error isolation downstream is a different concern: tolerating
-    a transient 5xx on ONE commit's status is best-effort; losing the
-    commit list itself means we don't even know which commits to try.)
+    Raises ApiError on non-2xx OR on unexpected response shape. The
+    branch-level caller soft-skips this tick because the next scheduled
+    tick can safely retry the listing. Per-SHA status/write errors remain
+    separate and must not be mislabeled as commit-list outages.
     """
     _, body = api(
         "GET",
@@ -656,7 +658,27 @@ def reap_branch(
       - compensated_per_sha: {<sha_full>: [<context>, ...]} — only
         SHAs that actually got at least one compensation are included
     """
-    shas = list_recent_commit_shas(branch, limit)
+    try:
+        shas = list_recent_commit_shas(branch, limit)
+    except ApiError as e:
+        print(
+            "::warning::status-reaper skipped this tick because the "
+            f"commit list could not be read after retries: {e}"
+        )
+        return {
+            "scanned_shas": 0,
+            "compensated": 0,
+            "preserved_real_push": 0,
+            "preserved_unknown": 0,
+            "preserved_non_failure": 0,
+            "preserved_non_push_suffix": 0,
+            "preserved_unparseable": 0,
+            "compensated_pr_shadowed_by_push_success": 0,
+            "preserved_pr_without_push_success": 0,
+            "compensated_per_sha": {},
+            "skipped": True,
+            "skip_reason": "commit-list-api-error",
+        }
 
     aggregate: dict[str, Any] = {
         "scanned_shas": 0,

.gitea/scripts/tests/test_gitea_merge_queue.py

@@ -85,7 +85,10 @@ def test_pr_needs_update_when_base_sha_absent_from_commits():
 
 def test_merge_decision_requires_main_green_pr_green_and_current_base():
     required = ["CI / all-required (pull_request)"]
-    main_status = {"state": "success", "statuses": []}
+    main_status = {
+        "state": "success",
+        "statuses": [{"context": "CI / all-required (push)", "status": "success"}],
+    }
     pr_status = {
         "state": "success",
         "statuses": [{"context": "CI / all-required (pull_request)", "status": "success"}],
@@ -104,7 +107,10 @@ def test_merge_decision_requires_main_green_pr_green_and_current_base():
 
 def test_merge_decision_updates_stale_pr_before_merge():
     decision = mq.evaluate_merge_readiness(
-        main_status={"state": "success", "statuses": []},
+        main_status={
+            "state": "success",
+            "statuses": [{"context": "CI / all-required (push)", "status": "success"}],
+        },
         pr_status={"state": "success", "statuses": [{"context": "CI / all-required (pull_request)", "status": "success"}]},
         required_contexts=["CI / all-required (pull_request)"],
         pr_has_current_base=False,

.gitea/scripts/tests/test_sop_checklist.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
-# Unit tests for sop-checklist-gate.py
+# Unit tests for sop-checklist.py
 #
-# Run:  python3 .gitea/scripts/tests/test_sop_checklist_gate.py
-#   or:  pytest .gitea/scripts/tests/test_sop_checklist_gate.py
+# Run:  python3 .gitea/scripts/tests/test_sop_checklist.py
+#   or:  pytest .gitea/scripts/tests/test_sop_checklist.py
 #
 # RFC#351 Step 2 of 6 — implementation MVP. Tests cover:
 #   - slug normalization (the 4 example variants in the script header)
@@ -33,7 +33,7 @@ sys.path.insert(0, PARENT)
 import importlib.util  # noqa: E402
 
 _spec = importlib.util.spec_from_file_location(
-    "sop_checklist_gate", os.path.join(PARENT, "sop-checklist-gate.py")
+    "sop_checklist", os.path.join(PARENT, "sop-checklist.py")
 )
 sop = importlib.util.module_from_spec(_spec)
 _spec.loader.exec_module(sop)  # type: ignore[union-attr]
@@ -134,18 +134,22 @@ class TestParseDirectives(unittest.TestCase):
     def setUp(self):
         self.aliases = _numeric_aliases()
 
+    def parse_ack_revoke(self, body):
+        directives, na_directives = sop.parse_directives(body, self.aliases)
+        self.assertEqual(na_directives, [])
+        return directives
+
     def test_simple_ack(self):
-        d = sop.parse_directives("/sop-ack comprehensive-testing", self.aliases)
+        d = self.parse_ack_revoke("/sop-ack comprehensive-testing")
         self.assertEqual(d, [("sop-ack", "comprehensive-testing", "")])
 
     def test_simple_revoke(self):
-        d = sop.parse_directives("/sop-revoke staging-smoke", self.aliases)
+        d = self.parse_ack_revoke("/sop-revoke staging-smoke")
         self.assertEqual(d, [("sop-revoke", "staging-smoke", "")])
 
     def test_ack_with_note(self):
-        d = sop.parse_directives(
-            "/sop-ack comprehensive-testing LGTM the test covers all edge cases",
-            self.aliases,
+        d = self.parse_ack_revoke(
+            "/sop-ack comprehensive-testing LGTM the test covers all edge cases"
         )
         self.assertEqual(len(d), 1)
         self.assertEqual(d[0][0], "sop-ack")
@@ -153,13 +157,12 @@ class TestParseDirectives(unittest.TestCase):
         self.assertIn("LGTM", d[0][2])
 
     def test_numeric_shorthand(self):
-        d = sop.parse_directives("/sop-ack 1", self.aliases)
+        d = self.parse_ack_revoke("/sop-ack 1")
         self.assertEqual(d, [("sop-ack", "comprehensive-testing", "")])
 
     def test_revoke_with_reason(self):
-        d = sop.parse_directives(
-            "/sop-revoke comprehensive-testing realized the e2e was mocking the DB",
-            self.aliases,
+        d = self.parse_ack_revoke(
+            "/sop-revoke comprehensive-testing realized the e2e was mocking the DB"
         )
         self.assertEqual(d[0][0], "sop-revoke")
         self.assertEqual(d[0][1], "comprehensive-testing")
@@ -171,7 +174,7 @@ class TestParseDirectives(unittest.TestCase):
             "/sop-ack comprehensive-testing\n"
             "Will follow up on the doc nit separately."
         )
-        d = sop.parse_directives(body, self.aliases)
+        d = self.parse_ack_revoke(body)
         self.assertEqual(len(d), 1)
         self.assertEqual(d[0][1], "comprehensive-testing")
 
@@ -180,7 +183,7 @@ class TestParseDirectives(unittest.TestCase):
             "/sop-ack comprehensive-testing\n"
             "/sop-ack local-postgres-e2e\n"
         )
-        d = sop.parse_directives(body, self.aliases)
+        d = self.parse_ack_revoke(body)
         self.assertEqual(len(d), 2)
         slugs = {x[1] for x in d}
         self.assertEqual(slugs, {"comprehensive-testing", "local-postgres-e2e"})
@@ -189,21 +192,21 @@ class TestParseDirectives(unittest.TestCase):
         # A directive embedded mid-line is not honored (prevents review
         # comments like "to /sop-ack you need..." from acting as acks).
         body = "If you want to /sop-ack comprehensive-testing reply in this thread"
-        d = sop.parse_directives(body, self.aliases)
+        d = self.parse_ack_revoke(body)
         self.assertEqual(d, [])
 
     def test_leading_whitespace_allowed(self):
         body = "  /sop-ack comprehensive-testing"
-        d = sop.parse_directives(body, self.aliases)
+        d = self.parse_ack_revoke(body)
         self.assertEqual(len(d), 1)
 
     def test_empty_body(self):
-        self.assertEqual(sop.parse_directives("", self.aliases), [])
-        self.assertEqual(sop.parse_directives(None, self.aliases), [])
+        self.assertEqual(sop.parse_directives("", self.aliases), ([], []))
+        self.assertEqual(sop.parse_directives(None, self.aliases), ([], []))
 
     def test_normalization_applied(self):
         # /sop-ack Comprehensive_Testing → canonical comprehensive-testing
-        d = sop.parse_directives("/sop-ack Comprehensive_Testing", self.aliases)
+        d = self.parse_ack_revoke("/sop-ack Comprehensive_Testing")
         self.assertEqual(d[0][1], "comprehensive-testing")
 
 

.gitea/scripts/tests/test_sop_tier_refire.sh

@@ -32,6 +32,7 @@ THIS_DIR="$(cd "$(dirname "$0")" && pwd)"
 SCRIPT_DIR="$(cd "$THIS_DIR/.." && pwd)"
 WORKFLOW_DIR="$(cd "$THIS_DIR/../../workflows" && pwd)"
 WORKFLOW="$WORKFLOW_DIR/sop-tier-refire.yml"
+DISPATCH_WORKFLOW="$WORKFLOW_DIR/review-refire-comments.yml"
 SCRIPT="$SCRIPT_DIR/sop-tier-refire.sh"
 
 PASS=0
@@ -87,6 +88,7 @@ assert_file_exists() {
 echo
 echo "== existence =="
 assert_file_exists "workflow file exists"  "$WORKFLOW"
+assert_file_exists "dispatcher workflow file exists" "$DISPATCH_WORKFLOW"
 assert_file_exists "script file exists"    "$SCRIPT"
 if [ "$FAIL" -gt 0 ]; then
   echo
@@ -104,29 +106,43 @@ echo "== T6/T7 workflow yaml =="
 PARSE_OUT=$(python3 -c 'import sys,yaml;yaml.safe_load(open(sys.argv[1]).read());print("ok")' "$WORKFLOW" 2>&1 || true)
 assert_eq "T7 workflow parses as YAML" "ok" "$PARSE_OUT"
 
-# Three required gates in the `if:` expression
+# The old per-workflow issue_comment listener caused queue storms because
+# Gitea queues jobs before evaluating job-level `if:`. The script remains,
+# but comment-triggered refires route through the single dispatcher.
 WORKFLOW_CONTENT=$(cat "$WORKFLOW")
-assert_contains "T6a workflow if: contains author_association gate" \
-  "github.event.comment.author_association" "$WORKFLOW_CONTENT"
-assert_contains "T6b workflow if: gates on MEMBER/OWNER/COLLABORATOR" \
-  '["MEMBER","OWNER","COLLABORATOR"]' "$WORKFLOW_CONTENT"
-assert_contains "T6c workflow if: contains slash-command trigger" \
-  "/refire-tier-check" "$WORKFLOW_CONTENT"
-assert_contains "T6d workflow if: gates on PR-not-issue" \
-  "github.event.issue.pull_request" "$WORKFLOW_CONTENT"
-assert_contains "T6e workflow listens on issue_comment" \
-  "issue_comment" "$WORKFLOW_CONTENT"
-assert_contains "T6f workflow requests statuses:write permission" \
-  "statuses: write" "$WORKFLOW_CONTENT"
-# Does NOT check out PR HEAD (security)
-if grep -q 'ref: \${{ github.event.pull_request.head' "$WORKFLOW"; then
-  echo "  FAIL  T6g workflow MUST NOT check out PR head (security)"
+if printf '%s' "$WORKFLOW_CONTENT" | grep -q '^  issue_comment:'; then
+  echo "  FAIL  T6a manual fallback workflow must not listen on issue_comment"
   FAIL=$((FAIL + 1))
-  FAILED_TESTS="${FAILED_TESTS} T6g"
+  FAILED_TESTS="${FAILED_TESTS} T6a"
 else
-  echo "  PASS  T6g workflow does not check out PR head"
+  echo "  PASS  T6a manual fallback workflow does not listen on issue_comment"
   PASS=$((PASS + 1))
 fi
+assert_contains "T6b workflow exposes workflow_dispatch" \
+  "workflow_dispatch" "$WORKFLOW_CONTENT"
+assert_contains "T6c workflow documents unsupported manual inputs" \
+  "workflow_dispatch inputs" "$WORKFLOW_CONTENT"
+# Does NOT check out PR HEAD (security)
+if grep -q 'ref: \${{ github.event.pull_request.head' "$WORKFLOW"; then
+  echo "  FAIL  T6d workflow MUST NOT check out PR head (security)"
+  FAIL=$((FAIL + 1))
+  FAILED_TESTS="${FAILED_TESTS} T6d"
+else
+  echo "  PASS  T6d workflow does not check out PR head"
+  PASS=$((PASS + 1))
+fi
+
+DISPATCH_PARSE_OUT=$(python3 -c 'import sys,yaml;yaml.safe_load(open(sys.argv[1]).read());print("ok")' "$DISPATCH_WORKFLOW" 2>&1 || true)
+assert_eq "T6e dispatcher workflow parses as YAML" "ok" "$DISPATCH_PARSE_OUT"
+DISPATCH_CONTENT=$(cat "$DISPATCH_WORKFLOW")
+assert_contains "T6f dispatcher listens on issue_comment" \
+  "issue_comment" "$DISPATCH_CONTENT"
+assert_contains "T6g dispatcher handles /qa-recheck" \
+  "/qa-recheck" "$DISPATCH_CONTENT"
+assert_contains "T6h dispatcher handles /security-recheck" \
+  "/security-recheck" "$DISPATCH_CONTENT"
+assert_contains "T6i dispatcher handles /refire-tier-check" \
+  "/refire-tier-check" "$DISPATCH_CONTENT"
 
 # T1-T5 — script behavior against a local Gitea-fixture
 echo

.gitea/sop-checklist-config.yaml

@@ -107,3 +107,39 @@ items:
     description: >-
       List of feedback memories applicable to this change. Ack from
       any engineer who has the same memory access.
+
+# N/A gate declarations (RFC#324 §N/A follow-up).
+# PRs where a gate genuinely does not apply (e.g., pure-infra with no
+# qa surface, or docs-only) can be declared N/A by a non-author peer
+# who is in one of the gate's required_teams. The sop-checklist
+# posts a `sop-checklist / na-declarations (pull_request)` status that
+# review-check.sh reads to skip the Gitea-APPROVE requirement.
+#
+# Usage: any PR commenter (peer) posts:
+#   /sop-n/a qa-review  <reason>
+#   /sop-n/a security-review  <reason>
+#
+# Slash commands:
+#   /sop-n/a <gate> [reason] — declare gate N/A (most-recent per-user wins)
+#   /sop-revoke <gate>      — revoke prior N/A declaration for that gate
+#
+# Gate names must match the context strings used by review-check.sh:
+#   qa-review      → qa-review / approved (<event>)        [TEAM_ID=20]
+#   security-review → security-review / approved (<event>)  [TEAM_ID=21]
+#
+# required_teams: OR semantics — any team member can declare N/A.
+# Authors cannot self-declare N/A (enforced by gate script).
+n/a_gates:
+  qa-review:
+    required_teams: [qa, security, engineers]
+    description: >-
+      QA review N/A when this change has no qa surface (pure-infra,
+      tooling-only, revert, dependency-only). A qa/eng/security member
+      must post /sop-n/a qa-review to activate.
+
+  security-review:
+    required_teams: [security, managers, ceo]
+    description: >-
+      Security review N/A when this change has no security surface
+      (docs-only, pure-frontend, dependency-only). A security/owners
+      member must post /sop-n/a security-review to activate.

.gitea/workflows/ci-mcp-stdio-transport.yml

@@ -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

.gitea/workflows/ci.yml

@@ -107,16 +107,25 @@ jobs:
             echo "scripts=true" >> "$GITHUB_OUTPUT"
             exit 0
           fi
-          # Both .github/workflows/ci.yml AND .gitea/workflows/ci.yml count
-          # as "this workflow changed" — either edit should force-run every
-          # downstream job. The Gitea port follows the same shape as the
-          # GitHub original so behavior matches when triggered on either
-          # platform.
-          DIFF=$(git diff --name-only "$BASE" HEAD 2>/dev/null || echo ".gitea/workflows/ci.yml")
-          echo "platform=$(echo "$DIFF" | grep -qE '^workspace-server/|^\.gitea/workflows/ci\.yml$|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "canvas=$(echo "$DIFF" | grep -qE '^canvas/|^\.gitea/workflows/ci\.yml$|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "python=$(echo "$DIFF" | grep -qE '^workspace/|^\.gitea/workflows/ci\.yml$|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
-          echo "scripts=$(echo "$DIFF" | grep -qE '^tests/e2e/|^scripts/|^infra/scripts/|^\.gitea/workflows/ci\.yml$|^\.github/workflows/ci\.yml$' && echo true || echo false)" >> "$GITHUB_OUTPUT"
+          # Workflow-only edits are covered by the workflow lint family
+          # and by this workflow's always-present required jobs. Do not fan
+          # those edits out into Go/Canvas/Python/shellcheck work; the
+          # downstream jobs still emit their required contexts via no-op
+          # steps when their surface flag is false.
+          #
+          # If the diff itself cannot be trusted, fail open by running every
+          # surface instead of silently under-testing the PR.
+          if ! DIFF=$(git diff --name-only "$BASE" HEAD 2>/dev/null); then
+            echo "platform=true" >> "$GITHUB_OUTPUT"
+            echo "canvas=true" >> "$GITHUB_OUTPUT"
+            echo "python=true" >> "$GITHUB_OUTPUT"
+            echo "scripts=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          echo "platform=$(echo "$DIFF" | grep -qE '^workspace-server/' && echo true || echo false)" >> "$GITHUB_OUTPUT"
+          echo "canvas=$(echo "$DIFF" | grep -qE '^canvas/' && echo true || echo false)" >> "$GITHUB_OUTPUT"
+          echo "python=$(echo "$DIFF" | grep -qE '^workspace/' && echo true || echo false)" >> "$GITHUB_OUTPUT"
+          echo "scripts=$(echo "$DIFF" | grep -qE '^tests/e2e/|^scripts/|^infra/scripts/' && echo true || echo false)" >> "$GITHUB_OUTPUT"
 
   # Platform (Go) — Go build/vet/test/lint + coverage gates. The always-run
   # + per-step gating shape preserves the GitHub-side required-check name
@@ -124,59 +133,49 @@ jobs:
   # the name match works on PRs that don't touch workspace-server/).
   platform-build:
     name: Platform (Go)
-    needs: changes
     runs-on: ubuntu-latest
-    # mc#774 (interim): re-mask platform-build pending fix-forward. Phase 4
-    # (#656) flipped this to continue-on-error: false based on a Phase-3-masked
-    # "green on main 2026-05-12" — the prior continue-on-error: true had
-    # been hiding failing tests in workspace-server/internal/handlers/.
-    # Two distinct failure classes surfaced on 0e5152c3:
-    #   (1) 4x delegation_test.go (lines 1110/1176/1228/1271): helpers
-    #       expectExecuteDelegationBase/Success/Failed are missing sqlmock
-    #       expectations for queries production has issued since ~2026-04-21
-    #       (last_outbound_at UPDATE, lookupDeliveryMode/Runtime SELECTs,
-    #       a2a_receive INSERT activity_logs, recordLedgerStatus writes).
-    #       Halt cond #3 applies (regression > 7 days → broader sweep).
-    #   (2) 1x mcp_test.go:433 (TestMCPHandler_CommitMemory_GlobalScope_Blocked):
-    #       commit 7d1a189f (2026-05-10) hardened mcp.go to scrub err.Error()
-    #       from JSON-RPC responses (OFFSEC-001), but the test asserts the
-    #       error message contains "GLOBAL". Production-vs-test contract
-    #       collision — needs design call, not mock update.
-    # Time-boxed Option A (90 min) did not fit the cross-cutting scope.
-    # This is a sequenced revert→fix→reflip per
-    # feedback_strict_root_only_after_class_a emergency clause — NOT
-    # a permanent re-mask. Re-flip blocked on mc#774 fix-forward landing.
-    # Other 4 #656 flips (changes, canvas-build, shellcheck, python-lint)
-    # retain continue-on-error: false; only platform-build regresses.
-    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
-    continue-on-error: true  # mc#774 fix-forward in flight; re-flip when mc#774 lands (PR #669 → rebase after #709)
+    # mc#774 (closed 2026-05-14): Phase 4 flip of the platform-build job.
+    # Phase 4 (#656) originally flipped this to continue-on-error: false based on
+    # Phase-3-masked "green on main 2026-05-12". Two failure classes then surfaced:
+    #   (1) 4x delegation_test.go sqlmock gaps (PR #669 / #634 fix-forward, closed).
+    #   (2) TestMCPHandler_CommitMemory_GlobalScope_Blocked (mcp_test.go:433):
+    #       OFFSEC-001 hardening collided with test assertion; tracked in mc#762.
+    # Fix-forward for (1) landed in PR #669. The mc#762 gap (2) is a separate
+    # issue — it does NOT block this flip because the test is already wrapped in
+    # the diagnostic step with its own continue-on-error: true (line 203).
+    # Flip confirmed by CI / Platform (Go) status = success on main HEAD 363905d3.
+    continue-on-error: false
+    # Job-level ceiling. The go test step below runs with a per-step 10m timeout;
+    # this cap catches any step that leaks past that. Set well above 10m so
+    # the per-step timeout is the active constraint.
+    timeout-minutes: 15
     defaults:
       run:
         working-directory: workspace-server
     steps:
-      - if: needs.changes.outputs.platform != 'true'
+      - if: false
         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'
+      - if: always()
         uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
         with:
           go-version: 'stable'
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         run: go mod download
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         run: go build ./cmd/server
       # CLI (molecli) moved to standalone repo: git.moleculesai.app/molecule-ai/molecule-cli
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         run: go vet ./...
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         name: Install golangci-lint
         run: go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.12.2
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         name: Run golangci-lint
         run: $(go env GOPATH)/bin/golangci-lint run --timeout 3m ./...
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         name: Diagnostic — per-package verbose 60s
         run: |
           set +e
@@ -192,11 +191,15 @@ jobs:
           echo "::endgroup::"
         # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
         continue-on-error: true
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         name: Run tests with race detection and coverage
-        run: go test -race -coverprofile=coverage.out ./...
+        # Explicit timeout: cold runner cache causes OOM kills at ~4m39s on the
+        # full ./... suite with race detection + coverage. A 10m per-step timeout
+        # lets the suite complete on cold cache (~5-7m) while failing cleanly
+        # instead of OOM-killing. The job-level timeout (15m) is a backstop.
+        run: go test -race -timeout 10m -coverprofile=coverage.out ./...
 
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         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
@@ -210,7 +213,7 @@ jobs:
                    END {for (f in s) printf "%6.1f%%  %s\n", s[f]/c[f], f}' \
             | sort -n
 
-      - if: needs.changes.outputs.platform == 'true'
+      - if: always()
         name: Check coverage thresholds
         # Enforces two gates from #1823 Layer 1:
         #   1. Total floor (25% — ratchet plan in COVERAGE_FLOOR.md).
@@ -298,28 +301,28 @@ jobs:
   # siblings — verified empirically on PR #2314).
   canvas-build:
     name: Canvas (Next.js)
-    needs: changes
     runs-on: ubuntu-latest
+    timeout-minutes: 20
     # Phase 4 (RFC #219 §1): confirmed green on main 2026-05-12.
     continue-on-error: false
     defaults:
       run:
         working-directory: canvas
     steps:
-      - if: needs.changes.outputs.canvas != 'true'
+      - if: false
         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'
+      - if: always()
         uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.canvas == 'true'
+      - if: always()
         uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
         with:
           node-version: '22'
-      - if: needs.changes.outputs.canvas == 'true'
+      - if: always()
         run: rm -f package-lock.json && npm install
-      - if: needs.changes.outputs.canvas == 'true'
+      - if: always()
         run: npm run build
-      - if: needs.changes.outputs.canvas == 'true'
+      - if: always()
         name: Run tests with coverage
         # Coverage instrumentation is configured in canvas/vitest.config.ts
         # (provider: v8, reporters: text + html + json-summary). Step 2 of
@@ -328,7 +331,7 @@ jobs:
         # tracked in #1815) after the team sees what current coverage is.
         run: npx vitest run --coverage
       - name: Upload coverage summary as artifact
-        if: needs.changes.outputs.canvas == 'true' && always()
+        if: 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
@@ -345,16 +348,15 @@ jobs:
   # Shellcheck (E2E scripts) — required check, always runs.
   shellcheck:
     name: Shellcheck (E2E scripts)
-    needs: changes
     runs-on: ubuntu-latest
     # Phase 4 (RFC #219 §1): confirmed green on main 2026-05-12.
     continue-on-error: false
     steps:
-      - if: needs.changes.outputs.scripts != 'true'
+      - if: false
         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'
+      - if: always()
         uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.scripts == 'true'
+      - if: always()
         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
@@ -365,32 +367,66 @@ jobs:
           find tests/e2e infra/scripts -type f -name '*.sh' -print0 \
             | xargs -0 shellcheck --severity=warning
 
-      - if: needs.changes.outputs.scripts == 'true'
+      - if: always()
         name: Lint cleanup-trap hygiene (RFC #2873)
         run: bash tests/e2e/lint_cleanup_traps.sh
 
-      - if: needs.changes.outputs.scripts == 'true'
+      - if: always()
         name: Run E2E bash unit tests (no live infra)
         run: |
           bash tests/e2e/test_model_slug.sh
 
+      - if: always()
+        name: Test ECR promote-tenant-image script (mock-driven, no live infra)
+        # Covers scripts/promote-tenant-image.sh — the codified
+        # :staging-latest → :latest ECR promote + tenant fleet redeploy
+        # closing molecule-ai/molecule-core#660. 40 mock-driven cases
+        # exercise every exit path (preflight, snapshot, promote, redeploy
+        # 403→SSM-refresh, verify, rollback). No live AWS/CP/SSM calls.
+        run: |
+          bash scripts/test-promote-tenant-image.sh
+
+      - if: always()
+        name: Shellcheck promote-tenant-image script
+        # scripts/ is excluded from the bulk shellcheck pass above (legacy
+        # SC3040/SC3043 cleanup pending). Run shellcheck explicitly on
+        # the promote script + its test harness so regressions there are
+        # caught by the required check.
+        run: |
+          shellcheck --severity=warning \
+            scripts/promote-tenant-image.sh \
+            scripts/test-promote-tenant-image.sh
+
+  # mc#959 root-fix (sre)
+
   canvas-deploy-reminder:
     name: Canvas Deploy Reminder
     runs-on: ubuntu-latest
-    # mc#774: pre-existing continue-on-error mask; root-fix and remove, do not renew silently.
-    continue-on-error: true
+    # 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
+    # `if:` makes the gating explicit so the drift script sees it.
+    # Runs on both main and staging pushes; step exits 0 when not applicable.
+    if: ${{ github.ref == 'refs/heads/main' || github.ref == 'refs/heads/staging' }}
     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 }}
+          CANVAS_CHANGED: ${{ needs.changes.outputs.canvas }}
+          EVENT_NAME: ${{ github.event_name }}
+          REF_NAME: ${{ github.ref }}
           # github.server_url resolves via the workflow-level env override
           # to the Gitea instance, so the RUN_URL points at the Gitea run
           # page (not github.com). See feedback_act_runner_github_server_url.
           RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
         run: |
+          set -euo pipefail
+          if [ "$CANVAS_CHANGED" != "true" ] || [ "$EVENT_NAME" != "push" ] || [ "$REF_NAME" != "refs/heads/main" ]; then
+            echo "Canvas deploy reminder not applicable for event=$EVENT_NAME ref=$REF_NAME canvas_changed=$CANVAS_CHANGED."
+            exit 0
+          fi
+
           # Write body to a temp file — avoids backtick escaping in shell.
           cat > /tmp/deploy-reminder.md << 'BODY'
           ## Canvas build passed — deploy required
@@ -422,7 +458,6 @@ jobs:
   # Python Lint & Test — required check, always runs.
   python-lint:
     name: Python Lint & Test
-    needs: changes
     runs-on: ubuntu-latest
     # Phase 4 (RFC #219 §1): confirmed green on main 2026-05-12.
     continue-on-error: false
@@ -432,25 +467,25 @@ jobs:
       run:
         working-directory: workspace
     steps:
-      - if: needs.changes.outputs.python != 'true'
+      - if: false
         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'
+      - if: always()
         uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
-      - if: needs.changes.outputs.python == 'true'
+      - if: always()
         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'
+      - if: always()
         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'
+      - if: always()
         run: python -m pytest --tb=short
 
-      - if: needs.changes.outputs.python == 'true'
+      - if: always()
         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. See issue #2790 for full rationale.
@@ -515,85 +550,104 @@ jobs:
     # red silently merged through. See internal#286 for the three concrete
     # tonight-of-2026-05-11 incidents that prompted the emergency bump.
     #
-    # Three properties of this job each close a failure mode:
+    # This job deliberately has no `needs:`. Gitea 1.22/act_runner can mark a
+    # job-level `if: always()` + `needs:` sentinel as skipped before upstream
+    # jobs settle, leaving branch protection with a permanent pending
+    # `CI / all-required` context. Instead, this independent sentinel polls the
+    # required commit-status contexts for this SHA and fails if any fail, skip,
+    # or never emit.
     #
-    #  1. `if: always()` — runs even when an upstream fails. Without it the
-    #     sentinel is `skipped` and protection treats that as missing → merge
-    #     ungated.
+    # canvas-deploy-reminder is intentionally NOT included in all-required.needs.
+    # It is an informational main-push reminder, not a PR quality gate. Keeping
+    # it in this dependency list lets a skipped reminder skip the required
+    # sentinel before the `always()` guard can emit a branch-protection status.
     #
-    #  2. Assertion is `result == "success"` per dep, NOT `!= "failure"`.
-    #     A `skipped` upstream (job gated by `if:` evaluating false, matrix
-    #     entry that couldn't run) must NOT silently pass through.
-    #     `skipped`-as-green is exactly the failure mode this gate closes.
-    #
-    #  3. `needs:` is the canonical list of "what counts as required."
-    #     status_check_contexts will reference only `ci/all-required` (Step 5
-    #     follow-up — branch-protection PATCH is Owners-tier per
-    #     `feedback_never_admin_merge_bypass`, separate PR); a new job is
-    #     added simply by listing it in `needs:` here.
-    #     `.gitea/workflows/ci-required-drift.yml` files a [ci-drift] issue
-    #     hourly if this list diverges from status_check_contexts or from
-    #     audit-force-merge.yml's REQUIRED_CHECKS env (RFC §4 + §6).
-    #
-    # Excluded from `needs:`: `canvas-deploy-reminder` — gated by
-    # `if: ... github.event_name == 'push' && github.ref == 'refs/heads/main'`,
-    # so on PR events it's legitimately `skipped`. The drift detector
-    # explicitly excludes `github.event_name`-gated jobs from F1 (see
-    # `.gitea/scripts/ci-required-drift.py::ci_job_names`).
-    #
-    # Phase 3 (RFC #219 §1) safety: underlying build jobs carry
-    # continue-on-error: true so their failures are masked to null (2026-05-12: re-enabled mc#774 interim)
-    # (Gitea suppresses status reporting for CoE jobs). This sentinel
-    # runs with continue-on-error: false so it always reports its
-    # result to the API — without this, the required-status entry
-    # (CI / all-required (pull_request)) is never created, which
-    # blocks PR merges. When Phase 3 ends, flip underlying jobs to
-    # continue-on-error: false; this sentinel can then be flipped to
-    # continue-on-error: true if a Phase-4 regression requires it.
     continue-on-error: false
     runs-on: ubuntu-latest
-    timeout-minutes: 1
-    needs:
-      - changes
-      - platform-build
-      - canvas-build
-      - shellcheck
-      - python-lint
-    if: always()
+    timeout-minutes: 45
     steps:
-      - name: Assert every required dependency succeeded
+      - name: Wait for required CI contexts
+        env:
+          GITEA_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          API_ROOT: ${{ github.server_url }}/api/v1
+          REPOSITORY: ${{ github.repository }}
+          COMMIT_SHA: ${{ github.sha }}
+          EVENT_NAME: ${{ github.event_name }}
         run: |
           set -euo pipefail
-          # `needs.*.result` is one of: success | failure | cancelled | skipped | null.
-          # We assert success per dep (not != failure) — see RFC §2 reasoning above.
-          # Null results are skipped: they come from Phase 3 (continue-on-error: true
-          # suppresses status) or from jobs still in-flight. The sentinel succeeds
-          # rather than blocking PRs on Phase 3 noise.
-          results='${{ toJSON(needs) }}'
-          echo "$results"
-          echo "$results" | python3 -c '
-          import json, sys
-          ns = json.load(sys.stdin)
-          # Phase 3 masked: jobs with continue-on-error: true may report "failure"
-          # Remove when mc#774 handler test failures are resolved.
-          PHASE3_MASKED = {"platform-build"}
-          # Exclude null (Phase 3 suppressed / in-flight) from the bad list.
-          bad = [(k, v.get("result")) for k, v in ns.items()
-                 if v.get("result") not in ("success", None, "cancelled", "skipped") and k not in PHASE3_MASKED]
-          if bad:
-              print(f"FAIL: jobs not green:", file=sys.stderr)
-              for k, r in bad:
-                  print(f"  - {k}: {r}", file=sys.stderr)
-              sys.exit(1)
-          pending = [(k, v.get("result")) for k, v in ns.items()
-                     if v.get("result") is None]
-          cancelled = [(k, v.get("result")) for k, v in ns.items()
-                       if v.get("result") == "cancelled"]
-          if pending:
-              print(f"WARN: {len(pending)} job(s) still in-flight (result=null): " +
-                    ", ".join(k for k, _ in pending), file=sys.stderr)
-          if cancelled:
-              print(f"INFO: {len(cancelled)} job(s) masked by continue-on-error: " +
-                    ", ".join(k for k, _ in cancelled), file=sys.stderr)
-          print(f"OK: all {len(ns)} required jobs succeeded (or Phase-3 suppressed)")
-          '
+          python3 - <<'PY'
+          import json
+          import os
+          import sys
+          import time
+          import urllib.error
+          import urllib.request
+
+          token = os.environ["GITEA_TOKEN"]
+          api_root = os.environ["API_ROOT"].rstrip("/")
+          repo = os.environ["REPOSITORY"]
+          sha = os.environ["COMMIT_SHA"]
+          event = os.environ["EVENT_NAME"]
+          required = [
+              f"CI / Detect changes ({event})",
+              f"CI / Platform (Go) ({event})",
+              f"CI / Canvas (Next.js) ({event})",
+              f"CI / Shellcheck (E2E scripts) ({event})",
+              f"CI / Python Lint & Test ({event})",
+          ]
+          terminal_bad = {"failure", "error"}
+          deadline = time.time() + 40 * 60
+          last_summary = None
+
+          def fetch_statuses():
+              statuses = []
+              for page in range(1, 6):
+                  url = f"{api_root}/repos/{repo}/commits/{sha}/statuses?page={page}&limit=100"
+                  req = urllib.request.Request(url, headers={"Authorization": f"token {token}"})
+                  with urllib.request.urlopen(req, timeout=10) as resp:
+                      chunk = json.load(resp)
+                  if not chunk:
+                      break
+                  statuses.extend(chunk)
+              latest = {}
+              for item in statuses:
+                  ctx = item.get("context")
+                  if not ctx:
+                      continue
+                  prev = latest.get(ctx)
+                  if prev is None or (item.get("updated_at") or item.get("created_at") or "") >= (prev.get("updated_at") or prev.get("created_at") or ""):
+                      latest[ctx] = item
+              return latest
+
+          while True:
+              try:
+                  latest = fetch_statuses()
+              except (TimeoutError, OSError, urllib.error.URLError) as exc:
+                  if time.time() >= deadline:
+                      print(f"FAIL: status polling did not recover before deadline: {exc}", file=sys.stderr)
+                      sys.exit(1)
+                  print(f"WARN: status poll failed, retrying: {exc}", flush=True)
+                  time.sleep(15)
+                  continue
+              states = {ctx: (latest.get(ctx) or {}).get("status") or (latest.get(ctx) or {}).get("state") or "missing" for ctx in required}
+              summary = ", ".join(f"{ctx}={state}" for ctx, state in states.items())
+              if summary != last_summary:
+                  print(summary, flush=True)
+                  last_summary = summary
+              bad = {ctx: state for ctx, state in states.items() if state in terminal_bad}
+              if bad:
+                  print("FAIL: required CI context failed:", file=sys.stderr)
+                  for ctx, state in bad.items():
+                      desc = (latest.get(ctx) or {}).get("description") or ""
+                      print(f"  - {ctx}: {state} {desc}", file=sys.stderr)
+                  sys.exit(1)
+              if all(state == "success" for state in states.values()):
+                  print(f"OK: all {len(required)} required CI contexts succeeded")
+                  sys.exit(0)
+              if time.time() >= deadline:
+                  print("FAIL: timed out waiting for required CI contexts:", file=sys.stderr)
+                  for ctx, state in states.items():
+                      print(f"  - {ctx}: {state}", file=sys.stderr)
+                  sys.exit(1)
+              time.sleep(15)
+          PY

.gitea/workflows/e2e-api.yml

@@ -69,6 +69,13 @@ name: E2E API Smoke Test
 # 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 #1046 (fixed 2026-05-14): Stale platform-server from cancelled runs
+#   lingers on :8080 after "Stop platform" step is skipped (workflow cancelled
+#   before reaching line 335). Added a pre-start "Kill stale platform-server"
+#   step (line 286) that scans /proc for zombie platform-server processes
+#   and kills them before the port probe or bind. Makes the ephemeral port
+#   probe + start sequence deterministic.
+#
 # 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
@@ -283,6 +290,35 @@ jobs:
           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 (issue #1046)
+        if: needs.detect-changes.outputs.api == 'true'
+        run: |
+          # Concurrent runs on the same host-network act_runner can leave a
+          # zombie platform-server from a cancelled/timeout run. Cancelled
+          # runs never reach the "Stop platform" step (line 335), so the
+          # old process lingers. Kill it before the ephemeral port probe
+          # or start so the port is definitively free.
+          #
+          # /proc scan — works on any Linux without pkill/lsof/ss.
+          # comm field is truncated to 15 chars: "platform-serve" matches
+          # "platform-server". Verify with cmdline to avoid false positives.
+          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}: ${cmdline}"
+              kill "$kpid" 2>/dev/null || true
+              killed=$((killed + 1))
+            fi
+          done
+          if [ "$killed" -gt 0 ]; then
+            sleep 2
+            echo "Killed $killed stale process(es); port(s) released."
+          else
+            echo "No stale platform-server found."
+          fi
       - name: Start platform (background)
         if: needs.detect-changes.outputs.api == 'true'
         working-directory: workspace-server
@@ -346,3 +382,4 @@ jobs:
         run: |
           docker rm -f "$PG_CONTAINER" 2>/dev/null || true
           docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true
+

.gitea/workflows/e2e-chat.yml

@@ -0,0 +1,288 @@
+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.
+#
+# Architecture:
+#   1. Ephemeral Postgres + Redis (docker, unique container names)
+#   2. workspace-server built from source, started with
+#      MOLECULE_ENV=development (fail-open auth)
+#   3. canvas dev server (npm run dev) on :3000
+#   4. Playwright tests create workspaces via API, point them at an
+#      in-process echo runtime, and exercise the full send/receive
+#      round-trip through the browser.
+#
+# Parallel-safety: same pattern as e2e-api.yml — per-run container names
+# and ephemeral host ports so concurrent jobs on the host-network runner
+# don't collide.
+
+on:
+  push:
+    branches: [main, staging]
+  pull_request:
+    branches: [main, staging]
+
+concurrency:
+  group: e2e-chat-${{ github.event.pull_request.head.sha || github.sha }}
+  cancel-in-progress: false
+
+env:
+  GITHUB_SERVER_URL: https://git.moleculesai.app
+
+jobs:
+  # bp-exempt: helper job; real gate is E2E Chat / E2E Chat (pull_request)
+  detect-changes:
+    runs-on: ubuntu-latest
+    # 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
+    outputs:
+      chat: ${{ steps.decide.outputs.chat }}
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          fetch-depth: 0
+      - id: decide
+        run: |
+          BASE="${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 }}"
+          fi
+          if [ -z "$BASE" ] || echo "$BASE" | grep -qE '^0+$'; then
+            echo "chat=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          if ! git cat-file -e "$BASE" 2>/dev/null; then
+            git fetch --depth=1 origin "$BASE" 2>/dev/null || true
+          fi
+          if ! git cat-file -e "$BASE" 2>/dev/null; then
+            echo "chat=true" >> "$GITHUB_OUTPUT"
+            exit 0
+          fi
+          CHANGED=$(git diff --name-only "$BASE" HEAD)
+          if echo "$CHANGED" | grep -qE '^(canvas/|workspace-server/|\.gitea/workflows/e2e-chat\.yml$)'; then
+            echo "chat=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "chat=false" >> "$GITHUB_OUTPUT"
+          fi
+
+  # bp-required: pending #1142 — new E2E check; add to branch protection after 3 green runs.
+  e2e-chat:
+    needs: detect-changes
+    name: E2E Chat
+    runs-on: ubuntu-latest
+    # 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
+    timeout-minutes: 15
+    env:
+      PG_CONTAINER: pg-e2e-chat-${{ github.run_id }}-${{ github.run_attempt }}
+      REDIS_CONTAINER: redis-e2e-chat-${{ github.run_id }}-${{ github.run_attempt }}
+    steps:
+      - name: No-op pass (paths filter excluded this commit)
+        if: needs.detect-changes.outputs.chat != 'true'
+        run: |
+          echo "No canvas / workspace-server / workflow changes — E2E Chat gate satisfied without running tests."
+          echo "::notice::E2E Chat no-op pass (paths filter excluded this commit)."
+
+      - if: needs.detect-changes.outputs.chat == 'true'
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - if: needs.detect-changes.outputs.chat == 'true'
+        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
+        with:
+          go-version: 'stable'
+          cache: true
+          cache-dependency-path: workspace-server/go.sum
+
+      - if: needs.detect-changes.outputs.chat == 'true'
+        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
+        with:
+          node-version: '22'
+          cache: 'npm'
+          cache-dependency-path: canvas/package-lock.json
+
+      - name: Start Postgres (docker)
+        if: needs.detect-changes.outputs.chat == 'true'
+        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 $2; exit}')
+          if [ -z "$PG_PORT" ]; then
+            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 $PG_CONTAINER"
+            exit 1
+          fi
+          echo "PG_PORT=${PG_PORT}" >> "$GITHUB_ENV"
+          echo "DATABASE_URL=postgres://dev:dev@127.0.0.1:${PG_PORT}/molecule?sslmode=disable" >> "$GITHUB_ENV"
+          echo "E2E_DATABASE_URL=postgres://dev:dev@127.0.0.1:${PG_PORT}/molecule?sslmode=disable" >> "$GITHUB_ENV"
+          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 ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "::error::Postgres did not become ready in 30s"
+          exit 1
+
+      - name: Start Redis (docker)
+        if: needs.detect-changes.outputs.chat == 'true'
+        run: |
+          docker rm -f "$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 $REDIS_CONTAINER"
+            exit 1
+          fi
+          echo "REDIS_PORT=${REDIS_PORT}" >> "$GITHUB_ENV"
+          echo "REDIS_URL=redis://127.0.0.1:${REDIS_PORT}" >> "$GITHUB_ENV"
+          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 ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "::error::Redis did not become ready in 15s"
+          exit 1
+
+      - name: Build platform
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: workspace-server
+        run: go build -o platform-server ./cmd/server
+
+      - name: Pick platform port
+        if: needs.detect-changes.outputs.chat == 'true'
+        run: |
+          PLATFORM_PORT=$(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 "PLATFORM_PORT=${PLATFORM_PORT}" >> "$GITHUB_ENV"
+          echo "E2E_PLATFORM_URL=http://127.0.0.1:${PLATFORM_PORT}" >> "$GITHUB_ENV"
+          echo "Platform host port: ${PLATFORM_PORT}"
+
+      - name: Pick canvas port
+        if: needs.detect-changes.outputs.chat == 'true'
+        run: |
+          CANVAS_PORT=$(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 "CANVAS_PORT=${CANVAS_PORT}" >> "$GITHUB_ENV"
+          echo "Canvas host port: ${CANVAS_PORT}"
+
+      - name: Start platform (background)
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: workspace-server
+        run: |
+          export MOLECULE_ENV=development
+          export DATABASE_URL="${DATABASE_URL}"
+          export REDIS_URL="${REDIS_URL}"
+          export PORT="${PLATFORM_PORT}"
+          export CORS_ORIGINS="http://localhost:3000,http://localhost:3001,http://localhost:${CANVAS_PORT},http://127.0.0.1:${CANVAS_PORT}"
+          ./platform-server > platform.log 2>&1 &
+          echo $! > platform.pid
+
+      - name: Wait for /health
+        if: needs.detect-changes.outputs.chat == 'true'
+        run: |
+          for i in $(seq 1 30); do
+            if curl -sf "http://127.0.0.1:${PLATFORM_PORT}/health" > /dev/null; then
+              echo "Platform up after ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "::error::Platform did not become healthy in 30s"
+          cat workspace-server/platform.log || true
+          exit 1
+
+      - name: Install canvas dependencies
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: canvas
+        run: npm ci
+
+      - name: Install Playwright browsers
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: canvas
+        run: npx playwright install --with-deps chromium
+
+      - name: Start canvas dev server (background)
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: canvas
+        run: |
+          export NEXT_PUBLIC_PLATFORM_URL="http://127.0.0.1:${PLATFORM_PORT}"
+          export NEXT_PUBLIC_WS_URL="ws://127.0.0.1:${PLATFORM_PORT}/ws"
+          npx next dev --turbopack -p "${CANVAS_PORT}" > canvas.log 2>&1 &
+          echo $! > canvas.pid
+          for i in $(seq 1 30); do
+            if curl -sf "http://localhost:${CANVAS_PORT}" > /dev/null 2>&1; then
+              echo "Canvas up after ${i}s"
+              exit 0
+            fi
+            sleep 1
+          done
+          echo "::error::Canvas did not start in 30s"
+          cat canvas.log || true
+          exit 1
+
+      - name: Run Playwright E2E tests
+        if: needs.detect-changes.outputs.chat == 'true'
+        working-directory: canvas
+        run: |
+          export E2E_PLATFORM_URL="http://127.0.0.1:${PLATFORM_PORT}"
+          export E2E_DATABASE_URL="${DATABASE_URL}"
+          export PLAYWRIGHT_BASE_URL="http://localhost:${CANVAS_PORT}"
+          npx playwright test e2e/chat-desktop.spec.ts e2e/chat-mobile.spec.ts
+
+      - name: Dump platform log on failure
+        if: failure() && needs.detect-changes.outputs.chat == 'true'
+        run: cat workspace-server/platform.log || true
+
+      - name: Dump canvas log on failure
+        if: failure() && needs.detect-changes.outputs.chat == 'true'
+        run: cat canvas/canvas.log || true
+
+      - name: Upload Playwright report
+        if: failure() && needs.detect-changes.outputs.chat == 'true'
+        uses: actions/upload-artifact@v3.2.2
+        with:
+          name: playwright-report-chat
+          path: canvas/playwright-report/
+
+      - name: Stop canvas
+        if: always() && needs.detect-changes.outputs.chat == 'true'
+        run: |
+          if [ -f canvas/canvas.pid ]; then
+            kill "$(cat canvas/canvas.pid)" 2>/dev/null || true
+          fi
+
+      - name: Stop platform
+        if: always() && needs.detect-changes.outputs.chat == 'true'
+        run: |
+          if [ -f workspace-server/platform.pid ]; then
+            kill "$(cat workspace-server/platform.pid)" 2>/dev/null || true
+          fi
+
+      - name: Stop service containers
+        if: always() && needs.detect-changes.outputs.chat == 'true'
+        run: |
+          docker rm -f "$PG_CONTAINER" 2>/dev/null || true
+          docker rm -f "$REDIS_CONTAINER" 2>/dev/null || true

.gitea/workflows/e2e-peer-visibility.yml

@@ -0,0 +1,225 @@
+name: E2E Peer Visibility (literal MCP list_peers)
+
+# WHY A DEDICATED WORKFLOW (not folded into e2e-staging-saas.yml)
+# --------------------------------------------------------------
+# This is the systemic fix for a real trust failure. Hermes and OpenClaw
+# were reported "fleet-verified / cascade-complete" because the *proxy*
+# signals were green (registry registration + heartbeat for Hermes; model
+# round-trip 200 for OpenClaw). A freshly-provisioned workspace asked on
+# canvas "can you see your peers" actually FAILS:
+#   - Hermes: 401 on the molecule MCP `list_peers` call
+#   - OpenClaw: native `sessions_list` fallback, sees no platform peers
+# Tasks #142/#159 were even marked "completed" under this proxy flaw.
+#
+# A dedicated workflow (vs extending e2e-staging-saas.yml) because:
+#   - It must provision MULTIPLE distinct runtimes (hermes, openclaw,
+#     claude-code) in ONE org and assert each sees the others. The
+#     full-saas script is single-runtime-per-run (E2E_RUNTIME) and folding
+#     a multi-runtime matrix into it would conflate concerns and bloat its
+#     already-45-min run.
+#   - It needs its own concurrency group so it doesn't fight full-saas /
+#     canvas for the staging org-creation quota.
+#   - It needs an independent, non-required status-context name so it can
+#     be RED today (the in-flight Hermes-401 / OpenClaw-MCP-wiring fixes
+#     have not landed) WITHOUT wedging unrelated merges — and flipped to
+#     REQUIRED in one branch-protection edit once it goes green
+#     (flip-to-required checklist: molecule-core#1296).
+#
+# THE ASSERTION IS NOT A PROXY. The driving script
+# tests/e2e/test_peer_visibility_mcp_staging.sh issues the byte-for-byte
+# JSON-RPC `tools/call name=list_peers` envelope to `POST
+# /workspaces/:id/mcp` using each workspace's OWN bearer token, through
+# the real WorkspaceAuth + MCPRateLimiter middleware chain — the exact
+# call mcp_molecule_list_peers makes from a canvas agent. It does NOT
+# read a registry row, /health, the heartbeat table, or
+# GET /registry/:id/peers.
+#
+# HONEST GATE — NO continue-on-error. Per feedback_fix_root_not_symptom a
+# fake-green mask would defeat the entire purpose. This workflow goes red
+# on today's broken behavior and green only when the root-cause fixes
+# actually land. It is intentionally NOT in branch_protections — see PR
+# body for the required-vs-not decision + flip tracking issue.
+#
+# Gitea 1.22.6 / act_runner notes honored:
+#   - No cross-repo `uses:` (feedback_gitea_cross_repo_uses_blocked). The
+#     actions/checkout SHA is the one e2e-staging-canvas.yml already uses
+#     successfully (a mirrored SHA — see #1277/PR#1292 root-cause).
+#   - Per-SHA concurrency, not global (feedback_concurrency_group_per_sha).
+#   - Workflow-level GITHUB_SERVER_URL pinned
+#     (feedback_act_runner_github_server_url).
+#   - pr-validate posts a status under the same check name so a
+#     workflow-only PR is not silently statusless and the context is
+#     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).
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - 'workspace-server/internal/handlers/mcp.go'
+      - 'workspace-server/internal/handlers/mcp_tools.go'
+      - 'workspace-server/internal/middleware/**'
+      - 'workspace-server/internal/handlers/registry.go'
+      - 'workspace-server/internal/handlers/workspace.go'
+      - 'workspace/a2a_mcp_server.py'
+      - 'workspace/platform_tools/registry.py'
+      - 'tests/e2e/test_peer_visibility_mcp_staging.sh'
+      - '.gitea/workflows/e2e-peer-visibility.yml'
+  pull_request:
+    branches: [main]
+    paths:
+      - 'workspace-server/internal/handlers/mcp.go'
+      - 'workspace-server/internal/handlers/mcp_tools.go'
+      - 'workspace-server/internal/middleware/**'
+      - 'workspace-server/internal/handlers/registry.go'
+      - 'workspace-server/internal/handlers/workspace.go'
+      - 'workspace/a2a_mcp_server.py'
+      - 'workspace/platform_tools/registry.py'
+      - 'tests/e2e/test_peer_visibility_mcp_staging.sh'
+      - '.gitea/workflows/e2e-peer-visibility.yml'
+  workflow_dispatch:
+  schedule:
+    # 07:30 UTC daily — catches AMI / template-hermes / template-openclaw
+    # drift even on quiet days. Offset 30m from e2e-staging-saas (07:00)
+    # so the two don't collide on the staging org-creation quota.
+    - cron: '30 7 * * *'
+
+concurrency:
+  # Per-SHA (feedback_concurrency_group_per_sha). A single global group
+  # would let a queued staging/main push behind a PR run get cancelled,
+  # leaving any gate that reads "completed run at SHA" stuck.
+  group: e2e-peer-visibility-${{ github.event.pull_request.head.sha || github.sha }}
+  cancel-in-progress: false
+
+env:
+  GITHUB_SERVER_URL: https://git.moleculesai.app
+
+jobs:
+  # PR path: post a real status under the required-ready check name so a
+  # workflow-only PR is never silently statusless. The actual EC2 E2E is
+  # push/dispatch/cron only (30+ min). This is NOT a fake-green mask of
+  # the real assertion — it validates the driving script's bash syntax
+  # and inline-python so a broken test script fails at PR time.
+  pr-validate:
+    name: E2E Peer Visibility
+    runs-on: ubuntu-latest
+    if: github.event_name == 'pull_request'
+    timeout-minutes: 5
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - name: Validate driving script
+        run: |
+          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"
+          echo "main / workflow_dispatch / daily cron (30+ min EC2 boot)."
+
+  # 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.
+  peer-visibility:
+    name: E2E Peer Visibility
+    runs-on: ubuntu-latest
+    if: github.event_name != 'pull_request'
+    timeout-minutes: 60
+
+    env:
+      MOLECULE_CP_URL: https://staging-api.moleculesai.app
+      MOLECULE_ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_API_TOKEN }}
+      # LLM provider key so each runtime can authenticate at boot.
+      # Priority MiniMax → direct-Anthropic → OpenAI matches
+      # test_staging_full_saas.sh's secrets-injection chain.
+      E2E_MINIMAX_API_KEY: ${{ secrets.MOLECULE_STAGING_MINIMAX_API_KEY }}
+      E2E_ANTHROPIC_API_KEY: ${{ secrets.MOLECULE_STAGING_ANTHROPIC_API_KEY }}
+      E2E_OPENAI_API_KEY: ${{ secrets.MOLECULE_STAGING_OPENAI_API_KEY }}
+      E2E_RUN_ID: "${{ github.run_id }}-${{ github.run_attempt }}"
+      PV_RUNTIMES: "hermes openclaw claude-code"
+
+    steps:
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+
+      - name: Verify admin token present
+        run: |
+          if [ -z "$MOLECULE_ADMIN_TOKEN" ]; then
+            echo "::error::CP_STAGING_ADMIN_API_TOKEN secret not set (Railway staging CP_ADMIN_API_TOKEN)"
+            exit 2
+          fi
+          echo "Admin token present"
+
+      - name: Verify an LLM key present
+        run: |
+          if [ -z "${E2E_MINIMAX_API_KEY:-}" ] && [ -z "${E2E_ANTHROPIC_API_KEY:-}" ] && [ -z "${E2E_OPENAI_API_KEY:-}" ]; then
+            echo "::error::No LLM provider key set — workspaces fail at boot with 'No provider API key found'. Set MOLECULE_STAGING_MINIMAX_API_KEY (or ANTHROPIC / OPENAI)."
+            exit 2
+          fi
+          echo "LLM key present"
+
+      - name: CP staging health preflight
+        run: |
+          code=$(curl -sS -o /dev/null -w "%{http_code}" --max-time 10 "$MOLECULE_CP_URL/health")
+          if [ "$code" != "200" ]; then
+            echo "::error::Staging CP unhealthy (HTTP $code) — infra, not a workspace bug. Failing loud per feedback_fix_root_not_symptom."
+            exit 1
+          fi
+          echo "Staging CP healthy"
+
+      - name: Run fresh-provision peer-visibility E2E (literal MCP list_peers)
+        run: bash tests/e2e/test_peer_visibility_mcp_staging.sh
+
+      # Belt-and-braces scoped teardown: the script installs an EXIT/INT/
+      # TERM trap, but if the runner itself is cancelled the trap may not
+      # fire. This always() step deletes ONLY the e2e-pv-<run_id> org this
+      # run created — never a cluster-wide sweep
+      # (feedback_never_run_cluster_cleanup_tests_on_live_platform). The
+      # admin DELETE is idempotent so double-invoking is safe;
+      # sweep-stale-e2e-orgs is the final net (slug starts with 'e2e-').
+      - name: Teardown safety net (runs on cancel/failure)
+        if: always()
+        env:
+          ADMIN_TOKEN: ${{ secrets.CP_STAGING_ADMIN_API_TOKEN }}
+        run: |
+          set +e
+          orgs=$(curl -sS "$MOLECULE_CP_URL/cp/admin/orgs?limit=500" \
+            -H "Authorization: Bearer $ADMIN_TOKEN" 2>/dev/null \
+            | python3 -c "
+          import json, sys, os, datetime
+          run_id = os.environ.get('GITHUB_RUN_ID', '')
+          try:
+              d = json.load(sys.stdin)
+          except Exception:
+              print(''); sys.exit(0)
+          # ONLY sweep slugs from THIS run. e2e-pv-<YYYYMMDD>-<run_id>-...
+          # Sweep today AND yesterday's UTC date so a midnight-crossing run
+          # still matches its own slug (same bug class as the saas/canvas
+          # safety nets).
+          today = datetime.date.today()
+          yest = today - datetime.timedelta(days=1)
+          dates = (today.strftime('%Y%m%d'), yest.strftime('%Y%m%d'))
+          if run_id:
+              prefixes = tuple(f'e2e-pv-{dt}-{run_id}-' for dt in dates)
+          else:
+              prefixes = tuple(f'e2e-pv-{dt}-' for dt in dates)
+          orgs = d if isinstance(d, list) else d.get('orgs', [])
+          cands = [o['slug'] for o in orgs
+                   if any(o.get('slug','').startswith(p) for p in prefixes)
+                   and o.get('instance_status') not in ('purged',)]
+          print('\n'.join(cands))
+          " 2>/dev/null)
+          for slug in $orgs; do
+            echo "Safety-net teardown: $slug"
+            set +e
+            curl -sS -o /tmp/pv-cleanup.out -w "%{http_code}" \
+              -X DELETE "$MOLECULE_CP_URL/cp/admin/tenants/$slug" \
+              -H "Authorization: Bearer $ADMIN_TOKEN" \
+              -H "Content-Type: application/json" \
+              -d "{\"confirm\":\"$slug\"}" >/tmp/pv-cleanup.code
+            set -e
+            code=$(cat /tmp/pv-cleanup.code 2>/dev/null || echo "000")
+            if [ "$code" = "200" ] || [ "$code" = "204" ]; then
+              echo "[teardown] deleted $slug (HTTP $code)"
+            else
+              echo "::warning::pv teardown for $slug returned HTTP $code — sweep-stale-e2e-orgs will catch it within MAX_AGE_MINUTES. Body: $(head -c 300 /tmp/pv-cleanup.out 2>/dev/null)"
+            fi
+          done
+          exit 0

.gitea/workflows/gate-check-v3.yml

@@ -83,25 +83,41 @@ jobs:
           REPO: ${{ github.repository }}
         run: |
           set -euo pipefail
-          # Fetch all open PRs and run gate-check on each
-          # socket.setdefaulttimeout(15): defence-in-depth for missing SOP_TIER_CHECK_TOKEN.
-          # gate_check.py uses timeout=15 on every urlopen call; this catches the
-          # inline Python polling loop too (issue #603).
+          # Fetch all open PRs and run gate-check on each. This scheduled
+          # refresher is advisory; a transient Gitea list timeout must not turn
+          # main red. PR-specific gate-check runs still use normal failure
+          # semantics.
           pr_numbers=$(python3 <<'PY'
           import json
           import os
           import socket
+          import sys
+          import time
+          import urllib.error
           import urllib.request
 
-          socket.setdefaulttimeout(15)
+          socket.setdefaulttimeout(30)
           token = os.environ["GITEA_TOKEN"]
           repo = os.environ["REPO"]
-          req = urllib.request.Request(
-              f"https://git.moleculesai.app/api/v1/repos/{repo}/pulls?state=open&limit=100",
-              headers={"Authorization": f"token {token}", "Accept": "application/json"},
-          )
-          with urllib.request.urlopen(req) as r:
-              prs = json.loads(r.read())
+          url = f"https://git.moleculesai.app/api/v1/repos/{repo}/pulls?state=open&limit=100"
+          last_error = None
+          for attempt in range(1, 4):
+              req = urllib.request.Request(
+                  url,
+                  headers={"Authorization": f"token {token}", "Accept": "application/json"},
+              )
+              try:
+                  with urllib.request.urlopen(req, timeout=30) as r:
+                      prs = json.loads(r.read())
+                  break
+              except (TimeoutError, OSError, urllib.error.URLError, urllib.error.HTTPError) as exc:
+                  last_error = exc
+                  print(f"warning: PR list fetch attempt {attempt}/3 failed: {exc}", file=sys.stderr)
+                  if attempt < 3:
+                      time.sleep(2 * attempt)
+          else:
+              print(f"warning: skipped scheduled gate-check refresh; failed to list open PRs after 3 attempts: {last_error}", file=sys.stderr)
+              raise SystemExit(0)
           for pr in prs:
               print(pr["number"])
           PY

.gitea/workflows/gitea-merge-queue.yml

@@ -48,4 +48,9 @@ jobs:
           REQUIRED_CONTEXTS: >-
             CI / all-required (pull_request),
             sop-checklist / all-items-acked (pull_request)
+          # Push-side required contexts. Checking CI / all-required (push)
+          # 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.
+          PUSH_REQUIRED_CONTEXTS: CI / all-required (push)
         run: python3 .gitea/scripts/gitea-merge-queue.py

.gitea/workflows/handlers-postgres-integration.yml

@@ -86,22 +86,33 @@ jobs:
     steps:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
-          fetch-depth: 0
+          # A full-history checkout can exceed the runner's quiet/startup
+          # window before the path filter emits logs. Fetch the common push
+          # case cheaply; the script below fetches the exact BASE SHA if it is
+          # not present in the shallow checkout.
+          fetch-depth: 2
       - id: filter
         # Inline replacement for dorny/paths-filter — see e2e-api.yml.
         run: |
-          BASE="${GITHUB_BASE_REF:-${{ github.event.before }}}"
+          # Gitea Actions evaluates github.event.before to empty string in shell
+          # scripts. Use GITHUB_EVENT_BEFORE shell env var instead (Gitea
+          # correctly populates it for push events). PR case uses template var.
+          BASE=""
           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" ]; then
+            BASE="$GITHUB_EVENT_BEFORE"
           fi
           if [ -z "$BASE" ] || echo "$BASE" | grep -qE '^0+$'; then
             echo "handlers=true" >> "$GITHUB_OUTPUT"
             exit 0
           fi
-          if ! git cat-file -e "$BASE" 2>/dev/null; then
+          # timeout 30 guards against the case where BASE points to a ref that
+          # git can resolve but cat-file hangs (rare on corrupted objects).
+          if ! timeout 30 git cat-file -e "$BASE" 2>/dev/null; then
             git fetch --depth=1 origin "$BASE" 2>/dev/null || true
           fi
-          if ! git cat-file -e "$BASE" 2>/dev/null; then
+          if ! timeout 30 git cat-file -e "$BASE" 2>/dev/null; then
             echo "handlers=true" >> "$GITHUB_OUTPUT"
             exit 0
           fi

.gitea/workflows/lint-continue-on-error-tracking.yml

@@ -93,7 +93,7 @@ jobs:
   lint:
     name: lint-continue-on-error-tracking
     runs-on: ubuntu-latest
-    timeout-minutes: 10
+    timeout-minutes: 20
     # Phase 3 (RFC #219 §1): surface masked defects without blocking
     # PRs. Pre-existing continue-on-error: true directives on main
     # all violate this lint at first — intentional. Flip to false

.gitea/workflows/publish-workspace-server-image.yml

@@ -37,12 +37,6 @@ name: publish-workspace-server-image
 on:
   push:
     branches: [main]
-    paths:
-      - 'workspace-server/**'
-      - 'canvas/**'
-      - 'manifest.json'
-      - 'scripts/**'
-      - '.gitea/workflows/publish-workspace-server-image.yml'
   workflow_dispatch:
 
 # No `concurrency:` block here. Gitea 1.22.6 can cancel queued runs despite
@@ -74,12 +68,14 @@ jobs:
           set -euo pipefail
           echo "::group::Docker daemon health check"
           echo "Runner: ${HOSTNAME:-unknown}"
-          docker info 2>&1 | head -5 || {
+          docker_info="$(docker info 2>&1)" || {
             echo "::error::Docker daemon is not accessible at /var/run/docker.sock"
             echo "::error::Runner: ${HOSTNAME:-unknown}"
+            printf '%s\n' "${docker_info}"
             echo "::error::Check: (1) daemon is running, (2) runner user is in docker group, (3) sock permissions are 660+"
             exit 1
           }
+          printf '%s\n' "${docker_info}" | sed -n '1,5p'
           echo "Docker daemon OK"
           echo "::endgroup::"
 

.gitea/workflows/qa-review.yml

@@ -9,10 +9,10 @@
 #   Triggers on:
 #     - `pull_request_target`: opened, synchronize, reopened
 #         → initial status posts when PR opens / re-pushes
-#     - `issue_comment`: /qa-recheck slash-command on the PR
-#         → manual re-fire after a QA reviewer clicks APPROVE
-#           (Gitea 1.22.6 doesn't re-fire on pull_request_review, per
-#           go-gitea/gitea#33700 + feedback_pull_request_review_no_refire)
+#     - comment refires are handled by `review-refire-comments.yml`
+#         → a single issue_comment dispatcher prevents every SOP/review
+#           comment from enqueueing separate qa/security/tier jobs on
+#           Gitea 1.22.6 before job-level `if:` can skip them.
 #   Workflow name = `qa-review` ; job name = `approved`.
 #   The job's own pass/fail conclusion publishes the status context
 #   `qa-review / approved (<event>)` — NO `POST /statuses` call → NO
@@ -85,8 +85,6 @@ name: qa-review
 on:
   pull_request_target:
     types: [opened, synchronize, reopened]
-  issue_comment:
-    types: [created]
 
 permissions:
   contents: read
@@ -97,16 +95,10 @@ jobs:
   approved:
     # Gate the job:
     #   - On pull_request_target events: always run.
-    #   - On issue_comment events: only when it's a PR comment and the body
-    #     contains the slash-command. NO privilege gate at the step level
-    #     (RFC#324 v1.3 §A1.1): a non-collaborator's /qa-recheck is fine
-    #     because the eval is read-only and idempotent — re-running it
-    #     just re-confirms whether a real team-member APPROVE exists.
+    # Comment-triggered refires live in review-refire-comments.yml. Keeping
+    # this workflow PR-only avoids comment-triggered queue storms.
     if: |
-      github.event_name == 'pull_request_target' ||
-      (github.event_name == 'issue_comment' &&
-       github.event.issue.pull_request != null &&
-       startsWith(github.event.comment.body, '/qa-recheck'))
+      github.event_name == 'pull_request_target'
     runs-on: ubuntu-latest
     steps:
       - name: Privilege check (A1.1 — INFORMATIONAL log only, NOT a gate)
@@ -120,7 +112,7 @@ jobs:
         # no comment.user.login so the step is a no-op skip there.
         if: github.event_name == 'issue_comment'
         env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
         run: |
           set -euo pipefail
           login="${{ github.event.comment.user.login }}"
@@ -151,7 +143,7 @@ jobs:
 
       - name: Evaluate qa-review
         env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
           GITEA_HOST: git.moleculesai.app
           REPO: ${{ github.repository }}
           # PR number lives in different places per event:

.gitea/workflows/redeploy-tenants-on-main.yml

@@ -9,19 +9,17 @@ name: redeploy-tenants-on-main
 #   - Workflow-level env.GITHUB_SERVER_URL pinned per
 #     feedback_act_runner_github_server_url.
 #   - `continue-on-error: true` on each job (RFC §1 contract).
-#   - ~~**Gitea workflow_run trigger limitation**~~ FIXED: replaced with
-#     push+paths filter per this PR. Gitea 1.22.6 does not support
-#     `workflow_run` (task #81). The push trigger fires on every
-#     commit to publish-workspace-server-image.yml which is the
-#     same signal (only successful runs commit to main).
+#   - Dropped unsupported `workflow_run` (task #81).
+#   - Later changed to manual-only after publish-workspace-server-image.yml
+#     gained an integrated ordered production deploy job.
 #
 
-# Auto-refresh prod tenant EC2s after every main merge.
+# Manual production tenant redeploy/rollback helper.
 #
-# 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.
+# Why this workflow is manual-only: publish-workspace-server-image now owns
+# the ordered build -> push -> production auto-deploy sequence in one workflow.
+# A separate push-triggered redeploy workflow races before the new ECR image
+# exists and can paint main red with a false deployment failure.
 #
 # This workflow closes the gap by calling the control-plane admin
 # endpoint that performs a canary-first, batched, health-gated rolling
@@ -34,36 +32,26 @@ name: redeploy-tenants-on-main
 # Gitea suspension migration. The staging-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.
+# Runtime ordering for automatic deploys now lives in
+# publish-workspace-server-image.yml:
+#   1. build-and-push creates new :staging-<sha> images in ECR.
+#   2. deploy-production waits for required push contexts on that SHA.
+#   3. deploy-production calls redeploy-fleet canary-first.
 #
-# 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.
+# Rollback path: set PROD_MANUAL_REDEPLOY_TARGET_TAG as a repo/org
+# variable or secret, run workflow_dispatch, then unset it after the
+# rollback. That calls redeploy-fleet with target_tag=<value>,
+# re-pulling the pinned image on every tenant.
 
 on:
-  push:
-    branches: [main]
-    paths:
-      - '.gitea/workflows/publish-workspace-server-image.yml'
   workflow_dispatch:
 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.
+# Serialize manual redeploys so two operator-triggered rollbacks do not
+# overlap and cause confusing per-tenant SSM state.
 #
 # NOTE: cancel-in-progress: false removed (Rule 7 fix). Gitea 1.22.6
 # cancels queued runs regardless of this setting, so it provides no
@@ -77,22 +65,17 @@ env:
   GITHUB_SERVER_URL: https://git.moleculesai.app
 
 jobs:
+  # bp-exempt: production redeploy is a side-effect workflow, not a merge gate.
   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.
-    # NOTE (Gitea port): workflow_dispatch trigger dropped; only the
-    # workflow_run path remains.
-    if: ${{ github.event.workflow_run.conclusion == 'success' }}
+    if: ${{ github.event_name == 'workflow_dispatch' }}
     runs-on: ubuntu-latest
     # 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
     timeout-minutes: 25
     env:
-      # Rule 9 fix: operational kill switch for auto-triggered deployments.
-      # Set repo variable or secret PROD_AUTO_DEPLOY_DISABLED=true to prevent
-      # this workflow from redeploying. Manual workflow_dispatch bypasses this.
+      # Rule 9 fix: keep the same operational kill switch surface as the
+      # integrated auto-deploy workflow.
       PROD_AUTO_DEPLOY_DISABLED: ${{ vars.PROD_AUTO_DEPLOY_DISABLED || secrets.PROD_AUTO_DEPLOY_DISABLED || '' }}
     steps:
       - name: Kill-switch guard
@@ -114,21 +97,16 @@ jobs:
         #      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 (staging-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.
+        #   2. Default → `staging-<short_head_sha>` for manual reruns from
+        #      the current default-branch SHA.
         env:
-          INPUT_TAG: ${{ inputs.target_tag }}
-          HEAD_SHA: ${{ github.event.workflow_run.head_sha || github.sha }}
+          PROD_MANUAL_REDEPLOY_TARGET_TAG: ${{ vars.PROD_MANUAL_REDEPLOY_TARGET_TAG || secrets.PROD_MANUAL_REDEPLOY_TARGET_TAG || '' }}
+          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: $INPUT_TAG"
+          if [ -n "${PROD_MANUAL_REDEPLOY_TARGET_TAG:-}" ]; then
+            echo "target_tag=$PROD_MANUAL_REDEPLOY_TARGET_TAG" >> "$GITHUB_OUTPUT"
+            echo "Using operator-pinned tag from PROD_MANUAL_REDEPLOY_TARGET_TAG."
           else
             SHORT="${HEAD_SHA:0:7}"
             echo "target_tag=staging-$SHORT" >> "$GITHUB_OUTPUT"
@@ -144,13 +122,26 @@ jobs:
           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 }}
+          CANARY_SLUG: ${{ vars.PROD_REDEPLOY_CANARY_SLUG || secrets.PROD_REDEPLOY_CANARY_SLUG || '' }}
+          SOAK_SECONDS: ${{ vars.PROD_REDEPLOY_SOAK_SECONDS || secrets.PROD_REDEPLOY_SOAK_SECONDS || '' }}
+          BATCH_SIZE: ${{ vars.PROD_REDEPLOY_BATCH_SIZE || secrets.PROD_REDEPLOY_BATCH_SIZE || '' }}
+          DRY_RUN: ${{ vars.PROD_REDEPLOY_DRY_RUN || secrets.PROD_REDEPLOY_DRY_RUN || '' }}
+          PROD_AUTO_DEPLOY_DISABLED: ${{ vars.PROD_AUTO_DEPLOY_DISABLED || secrets.PROD_AUTO_DEPLOY_DISABLED || '' }}
         run: |
           set -euo pipefail
 
+          case "${PROD_AUTO_DEPLOY_DISABLED,,}" in
+            1|true|yes|on)
+              echo "::notice::PROD_AUTO_DEPLOY_DISABLED is set; skipping production redeploy."
+              exit 0
+              ;;
+          esac
+
+          CANARY_SLUG="${CANARY_SLUG:-hongming}"
+          SOAK_SECONDS="${SOAK_SECONDS:-60}"
+          BATCH_SIZE="${BATCH_SIZE:-3}"
+          DRY_RUN="${DRY_RUN:-false}"
+
           if [ -z "${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."
@@ -172,7 +163,7 @@ jobs:
             }')
 
           echo "POST $CP_URL/cp/admin/tenants/redeploy-fleet"
-          echo "  body: $BODY"
+          echo "  target_tag=$TARGET_TAG canary=$CANARY_SLUG soak_seconds=$SOAK_SECONDS batch_size=$BATCH_SIZE dry_run=$DRY_RUN"
 
           HTTP_RESPONSE=$(mktemp)
           HTTP_CODE_FILE=$(mktemp)
@@ -261,13 +252,11 @@ jobs:
         # 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.
+        # When the redeploy is triggered manually with a specific tag
+        # (target_tag != "latest"), the expected SHA may not equal
+        # ${{ github.sha }}.
         env:
-          EXPECTED_SHA: ${{ github.event.workflow_run.head_sha || github.sha }}
+          EXPECTED_SHA: ${{ github.sha }}
           TARGET_TAG: ${{ steps.tag.outputs.target_tag }}
           # Tenant subdomain template — slugs from the response are
           # appended. Production CP issues `<slug>.moleculesai.app`;
@@ -281,10 +270,10 @@ jobs:
           if [ "$TARGET_TAG" != "latest" ] \
              && [ "$TARGET_TAG" != "$EXPECTED_SHA" ] \
              && [ "$TARGET_TAG" != "staging-$EXPECTED_SHORT" ]; then
-            # workflow_dispatch with a pinned tag that isn't the head
+            # Manual redeploy 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
+            # this context (would need to inspect the ECR
             # manifest, which is a follow-up). Failing-open here is
             # safe: the operator chose the tag deliberately.
             #

.gitea/workflows/review-refire-comments.yml

@@ -0,0 +1,113 @@
+# Consolidated comment dispatcher for manual review/tier refires.
+#
+# 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.
+
+name: review-refire-comments
+
+on:
+  issue_comment:
+    types: [created]
+
+permissions:
+  contents: read
+  pull-requests: read
+  statuses: write
+
+concurrency:
+  group: ${{ github.repository }}-${{ github.workflow }}-${{ github.event.issue.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  dispatch:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Classify comment
+        id: classify
+        env:
+          COMMENT_BODY: ${{ github.event.comment.body }}
+          IS_PR: ${{ github.event.issue.pull_request != null }}
+        run: |
+          set -euo pipefail
+          {
+            echo "run_qa=false"
+            echo "run_security=false"
+            echo "run_tier=false"
+          } >> "$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

.gitea/workflows/runtime-prbuild-compat.yml

@@ -66,19 +66,28 @@ jobs:
           # PR#372's ci.yml port used. Diffs against the PR base or the
           # previous push SHA, then matches against the wheel-relevant
           # path set.
-          BASE="${GITHUB_BASE_REF:-${{ github.event.before }}}"
-          if [ "${{ github.event_name }}" = "pull_request" ] && [ -n "${{ github.event.pull_request.base.sha }}" ]; then
+          #
+          # NOTE: Gitea Actions does not expose github.event.before as a
+          # shell environment variable. The ${{ github.event.before }} template
+          # expression works inside YAML run: blocks but is evaluated to an
+          # empty string for push events, making the ${VAR:-fallback} always
+          # use the fallback. Use GITHUB_EVENT_BEFORE instead — it IS set in
+          # the runner's shell environment for push events.
+          BASE=""
+          if [ "${{ github.event_name }}" = "pull_request" ]; then
             BASE="${{ github.event.pull_request.base.sha }}"
+          elif [ -n "$GITHUB_EVENT_BEFORE" ]; then
+            BASE="$GITHUB_EVENT_BEFORE"
           fi
           if [ -z "$BASE" ] || echo "$BASE" | grep -qE '^0+$'; then
             # New branch or no previous SHA: treat as wheel-relevant.
             echo "wheel=true" >> "$GITHUB_OUTPUT"
             exit 0
           fi
-          if ! git cat-file -e "$BASE" 2>/dev/null; then
+          if ! timeout 30 git cat-file -e "$BASE" 2>/dev/null; then
             git fetch --depth=1 origin "$BASE" 2>/dev/null || true
           fi
-          if ! git cat-file -e "$BASE" 2>/dev/null; then
+          if ! timeout 30 git cat-file -e "$BASE" 2>/dev/null; then
             echo "wheel=true" >> "$GITHUB_OUTPUT"
             exit 0
           fi

.gitea/workflows/security-review.yml

@@ -12,8 +12,6 @@ name: security-review
 on:
   pull_request_target:
     types: [opened, synchronize, reopened]
-  issue_comment:
-    types: [created]
 
 permissions:
   contents: read
@@ -22,13 +20,10 @@ permissions:
 jobs:
   # bp-exempt: PR security review bot signal; required merge state is enforced by CI / all-required.
   approved:
-    # See qa-review.yml header for full A1-α / A1.1 (v1.3 — informational
-    # log only, NOT a gate) / A4 / A5 design rationale.
+    # Comment-triggered refires live in review-refire-comments.yml. Keeping
+    # this workflow PR-only avoids comment-triggered queue storms.
     if: |
-      github.event_name == 'pull_request_target' ||
-      (github.event_name == 'issue_comment' &&
-       github.event.issue.pull_request != null &&
-       startsWith(github.event.comment.body, '/security-recheck'))
+      github.event_name == 'pull_request_target'
     runs-on: ubuntu-latest
     steps:
       - name: Privilege check (A1.1 — INFORMATIONAL log only, NOT a gate)
@@ -37,7 +32,7 @@ jobs:
         # so re-running on a non-collaborator comment is harmless.
         if: github.event_name == 'issue_comment'
         env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
         run: |
           set -euo pipefail
           login="${{ github.event.comment.user.login }}"
@@ -62,7 +57,7 @@ jobs:
 
       - name: Evaluate security-review
         env:
-          GITEA_TOKEN: ${{ secrets.RFC_324_TEAM_READ_TOKEN || secrets.GITHUB_TOKEN }}
+          GITEA_TOKEN: ${{ secrets.SOP_TIER_CHECK_TOKEN || secrets.GITHUB_TOKEN }}
           GITEA_HOST: git.moleculesai.app
           REPO: ${{ github.repository }}
           PR_NUMBER: ${{ github.event.pull_request.number || github.event.issue.number }}

.gitea/workflows/sop-checklist.yml

@@ -1,4 +1,4 @@
-# sop-checklist-gate — peer-ack merge gate for SOP-checklist items.
+# sop-checklist — peer-ack merge gate for SOP-checklist items.
 #
 # RFC#351 Step 2 of 6 (implementation MVP).
 #
@@ -65,7 +65,15 @@
 # membership, compute, post status). Re-running on any event is safe —
 # the new status overwrites the previous one for the same context.
 
-name: sop-checklist-gate
+name: sop-checklist
+
+# Cancel any in-progress runs for the same PR to prevent
+# stale runs from overwriting newer status contexts.
+concurrency:
+  group: ${{ github.repository }}-${{ github.workflow }}-${{ github.event.pull_request.number || github.event.issue.number || github.ref }}
+  cancel-in-progress: true
+
+# bp-required: yes  ← emits sop-checklist / all-items-acked (pull_request)
 
 on:
   pull_request_target:
@@ -83,7 +91,7 @@ permissions:
   statuses: write
 
 jobs:
-  gate:
+  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.
@@ -92,7 +100,8 @@ jobs:
       (github.event_name == 'issue_comment' &&
        github.event.issue.pull_request != null &&
        (contains(github.event.comment.body, '/sop-ack') ||
-        contains(github.event.comment.body, '/sop-revoke')))
+        contains(github.event.comment.body, '/sop-revoke') ||
+        contains(github.event.comment.body, '/sop-n/a')))
     runs-on: ubuntu-latest
     steps:
       - name: Check out BASE ref (trust boundary — never PR-head)
@@ -105,7 +114,7 @@ jobs:
           # qa-review.yml so the script source is always trusted.
           ref: ${{ github.event.repository.default_branch }}
 
-      - name: Run sop-checklist-gate
+      - name: Run sop-checklist
         env:
           GITEA_TOKEN: ${{ secrets.SOP_CHECKLIST_GATE_TOKEN || secrets.GITHUB_TOKEN }}
           PR_NUMBER: ${{ github.event.pull_request.number || github.event.issue.number }}
@@ -113,7 +122,7 @@ jobs:
           REPO_NAME: ${{ github.event.repository.name }}
         run: |
           set -euo pipefail
-          python3 .gitea/scripts/sop-checklist-gate.py \
+          python3 .gitea/scripts/sop-checklist.py \
             --owner "$OWNER" \
             --repo "$REPO_NAME" \
             --pr "$PR_NUMBER" \

.gitea/workflows/sop-tier-check.yml

@@ -61,6 +61,10 @@ on:
   pull_request_review:
     types: [submitted, dismissed, edited]
 
+concurrency:
+  group: ${{ github.repository }}-${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
 jobs:
   tier-check:
     runs-on: ubuntu-latest

.gitea/workflows/sop-tier-refire.yml

@@ -1,4 +1,4 @@
-# sop-tier-refire — issue_comment-triggered refire of sop-tier-check.
+# sop-tier-refire — manual fallback for sop-tier-check refire.
 #
 # Closes internal#292. Gitea 1.22.6 doesn't refire workflows on the
 # `pull_request_review` event (go-gitea/gitea#33700); the `sop-tier-check`
@@ -8,12 +8,12 @@
 # to merge is the admin force-merge path (audited via `audit-force-merge`
 # but the audit trail keeps growing; see `feedback_never_admin_merge_bypass`).
 #
-# Workaround pattern from `feedback_pull_request_review_no_refire`:
-# `issue_comment` events DO fire reliably on 1.22.6. When a repo
-# MEMBER/OWNER/COLLABORATOR comments `/refire-tier-check` on a PR, this
-# workflow re-runs the sop-tier-check logic and POSTs the resulting
-# status to the PR head SHA directly. No empty commit, no git history
-# bloat, no cascade re-fire of every other workflow on the PR.
+# Comment-triggered refires now live in `review-refire-comments.yml`. Gitea
+# queues issue_comment workflows before evaluating job-level `if:`, so having
+# qa-review, security-review, sop-checklist, and sop-tier-refire all subscribe
+# to every comment caused queue storms on SOP-heavy PRs. This workflow is a
+# non-automatic breadcrumb only; Gitea 1.22.6 does not support
+# workflow_dispatch inputs, so real refires must use `/refire-tier-check`.
 #
 # SECURITY MODEL:
 #
@@ -37,43 +37,16 @@
 # Rate-limit: a 1s pre-sleep + a "skip if status posted in last 30s"
 # guard prevents comment-spam from thrashing the status. See the script.
 
-name: sop-tier-check refire (issue_comment)
+name: sop-tier-check refire (manual)
 
 on:
-  issue_comment:
-    types: [created]
+  workflow_dispatch:
 
 jobs:
   refire:
-    # Three gates, all required:
-    #   - comment is on a PR (not a plain issue)
-    #   - commenter is MEMBER, OWNER, or COLLABORATOR
-    #   - comment body contains the slash-command trigger
-    if: |
-      github.event.issue.pull_request != null &&
-      contains(fromJson('["MEMBER","OWNER","COLLABORATOR"]'), github.event.comment.author_association) &&
-      contains(github.event.comment.body, '/refire-tier-check')
     runs-on: ubuntu-latest
-    permissions:
-      contents: read
-      pull-requests: read
-      statuses: write
     steps:
-      - name: Check out base branch (for the script)
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
-        with:
-          # Load the script from the default branch (main), matching the
-          # sop-tier-check.yml security model.
-          ref: ${{ github.event.repository.default_branch }}
-      - name: Re-evaluate sop-tier-check and POST status
-        env:
-          # Same org-level secret sop-tier-check.yml + audit-force-merge.yml use.
-          # Fallback to GITHUB_TOKEN with a clear error if missing.
-          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 }}
-          # Set to '1' for diagnostic per-API-call output. Off by default.
-          SOP_DEBUG: '0'
-        run: bash .gitea/scripts/sop-tier-refire.sh
+      - name: Explain supported refire path
+        run: |
+          echo "::error::Gitea 1.22.6 does not support workflow_dispatch inputs here; comment /refire-tier-check on the PR instead."
+          exit 1

.staging-trigger

@@ -1 +1 @@
-staging trigger
+staging trigger 2026-05-14T17:35:02Z

_ci_trigger.txt

@@ -0,0 +1 @@
+trigger

canvas/e2e/chat-desktop.spec.ts

@@ -0,0 +1,173 @@
+import { test, expect } from "@playwright/test";
+import { startEchoRuntime } from "./fixtures/echo-runtime";
+import { seedWorkspace, startHeartbeat, cleanupWorkspace } from "./fixtures/chat-seed";
+
+
+test.describe("Desktop ChatTab", () => {
+  let cleanup: () => Promise<void> = async () => {};
+  let workspaceId = "";
+  let workspaceName = "";
+
+  test.beforeAll(async () => {
+    const echo = await startEchoRuntime();
+    const ws = await seedWorkspace(echo.baseURL);
+    workspaceId = ws.id;
+    workspaceName = ws.name;
+    const stopHeartbeat = startHeartbeat(ws.id, ws.authToken);
+
+    cleanup = async () => {
+      stopHeartbeat();
+      await echo.stop();
+    };
+  });
+
+  test.afterAll(async () => {
+    await cleanupWorkspace(workspaceId);
+    await cleanup();
+  });
+
+  test.beforeEach(async ({ page }) => {
+    await page.setViewportSize({ width: 1280, height: 800 });
+    await page.goto("/");
+    await page.waitForSelector(".react-flow__node", { timeout: 10_000 });
+    // Dismiss onboarding guide if present.
+    const skipGuide = page.getByText("Skip guide");
+    if (await skipGuide.isVisible().catch(() => false)) {
+      await skipGuide.click();
+    }
+    // Click the workspace node by its exact name label.
+    await page.getByText(workspaceName, { exact: true }).first().click();
+    // Wait for the side panel chat tab to be clickable, then click it.
+    await page.locator('#tab-chat').click();
+    await page.waitForSelector("[data-testid='chat-panel']", { timeout: 5_000 });
+    // Wait for the workspace status to flip to online and the textarea to be enabled.
+    await expect(page.locator("textarea").first()).toBeEnabled({ timeout: 15_000 });
+  });
+
+  test("chat panel loads without error", async ({ page }) => {
+    const hasEmptyState = await page.getByText("Send a message to start chatting.").isVisible().catch(() => false);
+    const hasHistory = await page.locator("[data-testid='chat-panel']").locator("div").count() > 3;
+    expect(hasEmptyState || hasHistory).toBeTruthy();
+  });
+
+  test("send text message and receive echo response", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("What is the weather?");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("What is the weather?")).toBeVisible({ timeout: 5_000 });
+    await expect(page.getByText("Echo: What is the weather?")).toBeVisible({ timeout: 15_000 });
+  });
+
+  test("history persists across reload", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Persistence test");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Echo: Persistence test")).toBeVisible({ timeout: 15_000 });
+
+    await page.reload();
+    await page.waitForSelector(".react-flow__node", { timeout: 10_000 });
+    await page.getByText(workspaceName, { exact: true }).first().click();
+    await page.locator('#tab-chat').click();
+    await page.waitForSelector("[data-testid='chat-panel']", { timeout: 5_000 });
+    // Wait for the workspace status to flip to online and the textarea to be enabled.
+    await expect(page.locator("textarea").first()).toBeEnabled({ timeout: 15_000 });
+
+    await expect(page.getByText("Persistence test", { exact: true })).toBeVisible({ timeout: 5_000 });
+    await expect(page.getByText("Echo: Persistence test")).toBeVisible({ timeout: 5_000 });
+  });
+
+  test("file attachment round-trip", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Please read this file");
+
+    const fileInput = page.locator("[data-testid='chat-panel'] input[type='file']").first();
+    await fileInput.setInputFiles({
+      name: "test.txt",
+      mimeType: "text/plain",
+      buffer: Buffer.from("secret content abc123"),
+    });
+
+    await expect(page.getByText("test.txt")).toBeVisible({ timeout: 3_000 });
+
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Echo: Please read this file")).toBeVisible({ timeout: 15_000 });
+  });
+
+  test("activity log appears during send", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Trigger activity");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    // Activity log container should appear during the send flow.
+    await expect(page.locator("[data-testid='activity-log']").first()).toBeVisible({ timeout: 10_000 }).catch(() => {
+      // Activity log may not be present in all layouts.
+    });
+  });
+});
+
+test.describe("Desktop ChatTab — Markdown rendering", () => {
+  let cleanup: () => Promise<void> = async () => {};
+  let workspaceId = "";
+  let workspaceName = "";
+
+  test.beforeAll(async () => {
+    const echo = await startEchoRuntime();
+    const ws = await seedWorkspace(echo.baseURL);
+    workspaceId = ws.id;
+    workspaceName = ws.name;
+    const stopHeartbeat = startHeartbeat(ws.id, ws.authToken);
+
+    cleanup = async () => {
+      stopHeartbeat();
+      await echo.stop();
+    };
+  });
+
+  test.afterAll(async () => {
+    await cleanupWorkspace(workspaceId);
+    await cleanup();
+  });
+
+  test.beforeEach(async ({ page }) => {
+    await page.setViewportSize({ width: 1280, height: 800 });
+    await page.goto("/");
+    await page.waitForSelector(".react-flow__node", { timeout: 10_000 });
+    const skipGuide2 = page.getByText("Skip guide");
+    if (await skipGuide2.isVisible().catch(() => false)) {
+      await skipGuide2.click();
+    }
+    await page.getByText(workspaceName, { exact: true }).first().click();
+    await page.locator('#tab-chat').click();
+    await page.waitForSelector("[data-testid='chat-panel']", { timeout: 5_000 });
+    // Wait for the workspace status to flip to online and the textarea to be enabled.
+    await expect(page.locator("textarea").first()).toBeEnabled({ timeout: 15_000 });
+  });
+
+  test("code block renders <pre>", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("```js\nconst x = 1;\n```");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Echo: ```js")).toBeVisible({ timeout: 15_000 });
+
+    const pre = page.locator("pre").first();
+    await expect(pre).toBeVisible({ timeout: 5_000 });
+    await expect(pre).toContainText("const x = 1;");
+  });
+
+  test("table renders <table>", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("| A | B |\n|---|---|\n| 1 | 2 |");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Echo: | A | B |")).toBeVisible({ timeout: 15_000 });
+
+    const table = page.locator("table").first();
+    await expect(table).toBeVisible({ timeout: 5_000 });
+    await expect(table).toContainText("A");
+    await expect(table).toContainText("1");
+  });
+});

canvas/e2e/chat-mobile.spec.ts

@@ -0,0 +1,97 @@
+import { test, expect } from "@playwright/test";
+import { startEchoRuntime } from "./fixtures/echo-runtime";
+import { seedWorkspace, startHeartbeat, cleanupWorkspace } from "./fixtures/chat-seed";
+
+
+test.describe("MobileChat", () => {
+  let cleanup: () => Promise<void> = async () => {};
+  let workspaceId = "";
+
+  test.beforeAll(async () => {
+    const echo = await startEchoRuntime();
+    const ws = await seedWorkspace(echo.baseURL);
+    workspaceId = ws.id;
+    const stopHeartbeat = startHeartbeat(ws.id, ws.authToken);
+
+    cleanup = async () => {
+      stopHeartbeat();
+      await echo.stop();
+    };
+  });
+
+  test.afterAll(async () => {
+    await cleanupWorkspace(workspaceId);
+    await cleanup();
+  });
+
+  test.beforeEach(async ({ page }) => {
+    await page.setViewportSize({ width: 375, height: 812 });
+    // Navigate directly to the mobile chat view.
+    await page.goto(`/?m=chat&a=${workspaceId}`);
+    await page.waitForSelector("[data-testid='chat-panel']", { timeout: 10_000 });
+    // Wait for the workspace status to flip to online and the textarea to be enabled.
+    await expect(page.locator("textarea").first()).toBeEnabled({ timeout: 15_000 });
+    // Dismiss onboarding guide if present.
+    const skipGuide = page.getByText("Skip guide");
+    if (await skipGuide.isVisible().catch(() => false)) {
+      await skipGuide.click();
+    }
+  });
+
+  test("chat panel loads without error", async ({ page }) => {
+    const hasEmptyState = await page.getByText("Send a message to start chatting.").isVisible().catch(() => false);
+    const hasHistory = await page.locator("[data-testid='chat-panel']").locator("div").count() > 3;
+    expect(hasEmptyState || hasHistory).toBeTruthy();
+  });
+
+  test("send text message and receive echo response", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Mobile test message");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Mobile test message")).toBeVisible({ timeout: 5_000 });
+    await expect(page.getByText("Echo: Mobile test message")).toBeVisible({ timeout: 15_000 });
+  });
+
+  test("history persists across reload", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Mobile persistence");
+    await page.getByRole("button", { name: /Send/ }).first().click();
+
+    await expect(page.getByText("Echo: Mobile persistence")).toBeVisible({ timeout: 15_000 });
+
+    await page.reload();
+    await page.waitForSelector("[data-testid='chat-panel']", { timeout: 10_000 });
+
+    await expect(page.getByText("Mobile persistence", { exact: true })).toBeVisible({ timeout: 5_000 });
+    await expect(page.getByText("Echo: Mobile persistence")).toBeVisible({ timeout: 5_000 });
+  });
+
+  test("composer auto-grows with multi-line text", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    const initialHeight = await textarea.evaluate((el: HTMLElement) => el.offsetHeight);
+
+    await textarea.fill("Line 1\nLine 2\nLine 3\nLine 4\nLine 5");
+    await page.waitForTimeout(300);
+
+    const grownHeight = await textarea.evaluate((el: HTMLElement) => el.offsetHeight);
+    expect(grownHeight).toBeGreaterThan(initialHeight);
+  });
+
+  test("file attachment in mobile chat", async ({ page }) => {
+    const textarea = page.locator("textarea").first();
+    await textarea.fill("Mobile file test");
+
+    const fileInput = page.locator("[data-testid='chat-panel'] input[type='file']").first();
+    await fileInput.setInputFiles({
+      name: "mobile.txt",
+      mimeType: "text/plain",
+      buffer: Buffer.from("mobile secret"),
+    });
+
+    await expect(page.getByText("mobile.txt")).toBeVisible({ timeout: 3_000 });
+
+    await page.getByRole("button", { name: /Send/ }).first().click();
+    await expect(page.getByText("Echo: Mobile file test")).toBeVisible({ timeout: 15_000 });
+  });
+});

canvas/e2e/fixtures/chat-seed.ts

@@ -0,0 +1,187 @@
+/**
+ * E2E seed fixture for chat tests.
+ *
+ * Creates an external workspace via the workspace-server API, extracts the
+ * auto-minted auth token, then overrides the DB row so it appears "online"
+ * with an echo-runtime URL.  External runtime is used because the health
+ * sweep skips Docker checks for external workspaces; we keep the workspace
+ * alive with periodic heartbeats.
+ */
+
+import { randomUUID } from "node:crypto";
+
+const PLATFORM_URL = process.env.E2E_PLATFORM_URL ?? "http://localhost:8080";
+
+export interface SeededWorkspace {
+  id: string;
+  name: string;
+  agentURL: string;
+  authToken: string;
+}
+
+/**
+ * Create an external workspace and wire it to the echo runtime.
+ */
+export async function seedWorkspace(echoURL: string): Promise<SeededWorkspace> {
+  // 1. Create external workspace (no URL — platform will mint an auth token).
+  const runId = Math.random().toString(36).slice(2, 8);
+  const wsName = `Chat E2E Agent ${runId}`;
+  const createRes = await fetch(`${PLATFORM_URL}/workspaces`, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ name: wsName, tier: 1, external: true, runtime: "external" }),
+  });
+  if (!createRes.ok) {
+    const text = await createRes.text();
+    throw new Error(`Failed to create workspace: ${createRes.status} ${text}`);
+  }
+  const ws = (await createRes.json()) as {
+    id: string;
+    name: string;
+    connection?: { auth_token?: string };
+  };
+  const authToken = ws.connection?.auth_token;
+  if (!authToken) {
+    throw new Error("Workspace created but no auth_token returned");
+  }
+
+  // 2. Direct DB update: mark online + point url at echo runtime.
+  //    The platform blocks loopback URLs at the API layer (SSRF guard),
+  //    so we bypass via psql for local E2E.
+  const dbUrl = process.env.E2E_DATABASE_URL;
+  if (!dbUrl) {
+    throw new Error("E2E_DATABASE_URL must be set for DB seeding");
+  }
+  const pgRegex = /postgres:\/\/([^:]+):([^@]+)@([^:]+):(\d+)\/([^?]+)/;
+  const m = dbUrl.match(pgRegex);
+  if (!m) {
+    throw new Error(`Cannot parse E2E_DATABASE_URL: ${dbUrl}`);
+  }
+  const [, user, pass, host, port, db] = m;
+
+  // Pre-seed a platform_inbound_secret so chat file uploads don't trigger
+  // the lazy-heal 503 "retry in 30 s" path on first use.
+  const inboundSecret = Array.from({ length: 43 }, () =>
+    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_"[
+      Math.floor(Math.random() * 64)
+    ],
+  ).join("");
+
+  const psql = [
+    `PGPASSWORD=${pass} psql`,
+    `-h ${host} -p ${port} -U ${user} -d ${db}`,
+    `-c "UPDATE workspaces SET status = 'online', url = '${echoURL}', platform_inbound_secret = '${inboundSecret}' WHERE id = '${ws.id}'"`,
+  ].join(" ");
+
+  const { execSync } = await import("node:child_process");
+  try {
+    execSync(psql, { stdio: "pipe", timeout: 30_000 });
+  } catch (err) {
+    throw new Error(`DB update failed: ${err}`);
+  }
+
+  return { id: ws.id, name: wsName, agentURL: echoURL, authToken };
+}
+
+/**
+ * Start a heartbeat interval that keeps an external workspace alive.
+ * Returns a stop function.
+ */
+export function startHeartbeat(
+  workspaceId: string,
+  authToken: string,
+  intervalMs = 30_000,
+): () => void {
+  const send = () => {
+    fetch(`${PLATFORM_URL}/registry/heartbeat`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        Authorization: `Bearer ${authToken}`,
+      },
+      body: JSON.stringify({
+        workspace_id: workspaceId,
+        error_rate: 0,
+        sample_error: "",
+        active_tasks: 0,
+        current_task: "",
+        uptime_seconds: 0,
+      }),
+    }).catch(() => {});
+  };
+
+  // Send immediately so the first heartbeat lands before the stale sweep.
+  send();
+  const timer = setInterval(send, intervalMs);
+
+  return () => clearInterval(timer);
+}
+
+/**
+ * Seed chat-history rows for a workspace.
+ */
+export async function seedChatHistory(
+  workspaceId: string,
+  messages: Array<{ role: "user" | "agent"; content: string }>,
+): Promise<void> {
+  const dbUrl = process.env.E2E_DATABASE_URL;
+  if (!dbUrl) return;
+
+  const pgRegex = /postgres:\/\/([^:]+):([^@]+)@([^:]+):(\d+)\/([^?]+)/;
+  const m = dbUrl.match(pgRegex);
+  if (!m) return;
+  const [, user, pass, host, port, db] = m;
+
+  const values = messages
+    .map(
+      (msg, i) =>
+        `('${randomUUID()}', '${workspaceId}', '${msg.role}', '${msg.content.replace(/'/g, "''")}', NOW() - INTERVAL '${messages.length - i} seconds')`,
+    )
+    .join(",");
+
+  const sql = `INSERT INTO chat_messages (id, workspace_id, role, content, created_at) VALUES ${values};`;
+
+  const { execSync } = await import("node:child_process");
+  const psql = `PGPASSWORD=${pass} psql -h ${host} -p ${port} -U ${user} -d ${db} -c "${sql}"`;
+  execSync(psql, { stdio: "pipe", timeout: 10_000 });
+}
+
+/**
+ * Delete a seeded workspace row directly from the DB.
+ * Uses psql (same credentials as seedWorkspace) so we bypass any
+ * workspace-server side-effects (container stop, cascade cleanup, etc.)
+ * that can race or 500 on external workspaces.
+ */
+export async function cleanupWorkspace(workspaceId: string): Promise<void> {
+  const dbUrl = process.env.E2E_DATABASE_URL;
+  if (!dbUrl) return;
+
+  const pgRegex = /postgres:\/\/([^:]+):([^@]+)@([^:]+):(\d+)\/([^?]+)/;
+  const m = dbUrl.match(pgRegex);
+  if (!m) return;
+  const [, user, pass, host, port, db] = m;
+
+  const psql = `PGPASSWORD=${pass} psql -h ${host} -p ${port} -U ${user} -d ${db} -c "DELETE FROM workspaces WHERE id = '${workspaceId}'"`;
+
+  const { execSync } = await import("node:child_process");
+  try {
+    execSync(psql, { stdio: "pipe", timeout: 30_000 });
+  } catch {
+    // Best-effort cleanup; don't fail the test suite if the row is already gone.
+  }
+}
+
+/**
+ * Mint a workspace auth token so the canvas can make authenticated API
+ * calls (WorkspaceAuth middleware).
+ */
+export async function mintTestToken(workspaceId: string): Promise<string> {
+  const res = await fetch(
+    `${PLATFORM_URL}/admin/workspaces/${workspaceId}/test-token`,
+  );
+  if (!res.ok) {
+    throw new Error(`Failed to mint test token: ${res.status}`);
+  }
+  const data = (await res.json()) as { auth_token: string };
+  return data.auth_token;
+}

canvas/e2e/fixtures/echo-runtime.ts

@@ -0,0 +1,180 @@
+/**
+ * Minimal A2A echo runtime for E2E tests.
+ *
+ * Listens on an ephemeral port, receives A2A JSON-RPC `message/send`
+ * requests, and returns a response with the original text echoed back.
+ * Also implements the workspace-side chat upload ingest endpoint so
+ * file-attachment E2E can exercise the full upload → send → echo
+ * round-trip.
+ *
+ * Usage (inside test fixture):
+ *   const echo = await startEchoRuntime();
+ *   // ... seed workspace with agent_url pointing to echo.baseURL ...
+ *   echo.stop();
+ */
+
+import { createServer, type Server } from "node:http";
+
+export interface EchoRuntime {
+  baseURL: string;
+  stop: () => Promise<void>;
+  lastRequest: { method: string; text: string; files: unknown[] } | null;
+}
+
+/** Parse a minimal multipart body and extract the first file's name + content. */
+function parseMultipart(body: Buffer): { name: string; mimeType: string; content: Buffer } | null {
+  // Find the boundary line (first line starting with "--").
+  const str = body.toString("binary");
+  const firstDash = str.indexOf("--");
+  if (firstDash === -1) return null;
+  const eol = str.indexOf("\r\n", firstDash);
+  if (eol === -1) return null;
+  const boundary = str.slice(firstDash + 2, eol);
+  const boundaryMarker = "\r\n--" + boundary;
+
+  // Find the first part that has a filename in Content-Disposition.
+  let pos = eol + 2;
+  while (pos < str.length) {
+    const nextBoundary = str.indexOf(boundaryMarker, pos);
+    if (nextBoundary === -1) break;
+    const part = str.slice(pos, nextBoundary);
+
+    const cdMatch = part.match(/Content-Disposition:[^\r\n]*filename="([^"]+)"/i);
+    if (cdMatch) {
+      const name = cdMatch[1];
+      const ctMatch = part.match(/Content-Type:\s*([^\r\n]+)/i);
+      const mimeType = ctMatch ? ctMatch[1].trim() : "application/octet-stream";
+      // Body starts after the first double-CRLF in the part.
+      const bodyStart = part.indexOf("\r\n\r\n");
+      if (bodyStart !== -1) {
+        // Extract the raw bytes (not the string) so binary is safe.
+        const headerBytes = Buffer.byteLength(part.slice(0, bodyStart + 4), "binary");
+        const partStartInBody = Buffer.byteLength(str.slice(0, pos + bodyStart + 4), "binary");
+        const partEndInBody = Buffer.byteLength(str.slice(0, nextBoundary), "binary");
+        const content = body.subarray(partStartInBody, partEndInBody);
+        return { name, mimeType, content };
+      }
+    }
+    pos = nextBoundary + boundaryMarker.length;
+    // Skip trailing "--" (end marker) or CRLF.
+    if (str.slice(pos, pos + 2) === "--") break;
+    if (str.slice(pos, pos + 2) === "\r\n") pos += 2;
+  }
+  return null;
+}
+
+export async function startEchoRuntime(): Promise<EchoRuntime> {
+  let lastRequest: EchoRuntime["lastRequest"] = null;
+
+  const server = createServer((req, res) => {
+    // CORS: allow the canvas origin (localhost:3000) to call us.
+    res.setHeader("Access-Control-Allow-Origin", "*");
+    res.setHeader("Access-Control-Allow-Methods", "POST, GET, OPTIONS");
+    res.setHeader("Access-Control-Allow-Headers", "Content-Type, Authorization");
+
+    if (req.method === "OPTIONS") {
+      res.writeHead(204);
+      res.end();
+      return;
+    }
+
+    const url = req.url ?? "/";
+
+    // Workspace-side chat upload ingest (RFC #2312).
+    if (url === "/internal/chat/uploads/ingest" && req.method === "POST") {
+      const chunks: Buffer[] = [];
+      req.on("data", (chunk: Buffer) => chunks.push(chunk));
+      req.on("end", () => {
+        const body = Buffer.concat(chunks);
+        const file = parseMultipart(body);
+        if (!file) {
+          res.writeHead(400);
+          res.end(JSON.stringify({ error: "no files field" }));
+          return;
+        }
+        const sanitized = file.name.replace(/[^a-zA-Z0-9._\-]/g, "_").replace(/ /g, "_");
+        const prefix = Array.from({ length: 32 }, () =>
+          Math.floor(Math.random() * 16).toString(16),
+        ).join("");
+        const response = {
+          files: [
+            {
+              uri: `workspace:/workspace/.molecule/chat-uploads/${prefix}-${sanitized}`,
+              name: sanitized,
+              mimeType: file.mimeType,
+              size: file.content.length,
+            },
+          ],
+        };
+        res.setHeader("Content-Type", "application/json");
+        res.writeHead(200);
+        res.end(JSON.stringify(response));
+      });
+      return;
+    }
+
+    // Default: A2A JSON-RPC handler.
+    let body = "";
+    req.setEncoding("utf8");
+    req.on("data", (chunk: string) => {
+      body += chunk;
+    });
+    req.on("end", () => {
+      res.setHeader("Content-Type", "application/json");
+      try {
+        const rpc = JSON.parse(body);
+        const msg = rpc.params?.message;
+        const textParts =
+          msg?.parts
+            ?.filter((p: { kind?: string; text?: string }) => p.kind === "text")
+            .map((p: { text?: string }) => p.text)
+            .filter(Boolean) ?? [];
+        const fileParts =
+          msg?.parts?.filter((p: { kind?: string }) => p.kind === "file") ?? [];
+        const text = textParts.join("\n");
+
+        lastRequest = {
+          method: rpc.method ?? "unknown",
+          text,
+          files: fileParts,
+        };
+
+        const replyText = text
+          ? `Echo: ${text}`
+          : fileParts.length > 0
+            ? "Echo: received your file(s)."
+            : "Echo: hello";
+
+        const response = {
+          jsonrpc: "2.0",
+          id: rpc.id ?? null,
+          result: {
+            parts: [{ kind: "text", text: replyText }],
+          },
+        };
+
+        res.writeHead(200);
+        res.end(JSON.stringify(response));
+      } catch {
+        res.writeHead(400);
+        res.end(JSON.stringify({ error: "invalid json" }));
+      }
+    });
+  });
+
+  await new Promise<void>((resolve) => server.listen(0, "127.0.0.1", resolve));
+  const address = server.address();
+  const port = typeof address === "object" && address ? address.port : 0;
+  const baseURL = `http://127.0.0.1:${port}`;
+
+  return {
+    baseURL,
+    stop: () =>
+      new Promise((resolve) => {
+        server.close(() => resolve(undefined));
+      }),
+    get lastRequest() {
+      return lastRequest;
+    },
+  };
+}

canvas/playwright.config.ts

@@ -5,9 +5,10 @@ export default defineConfig({
   timeout: 30_000,
   expect: { timeout: 10_000 },
   fullyParallel: false,
+  workers: 1,
   retries: 0,
   use: {
-    baseURL: "http://localhost:3000",
+    baseURL: process.env.PLAYWRIGHT_BASE_URL || "http://localhost:3000",
     headless: true,
     screenshot: "only-on-failure",
   },

canvas/src/app/orgs/page.tsx

@@ -327,7 +327,7 @@ function OrgCTA({ org }: { org: Org }) {
     return (
       <a
         href={href}
-        className="rounded bg-emerald-600 px-4 py-2 text-sm font-medium text-white hover:bg-emerald-500"
+        className="rounded bg-emerald-700 px-4 py-2 text-sm font-medium text-white hover:bg-emerald-600"
       >
         Open
       </a>
@@ -337,7 +337,7 @@ function OrgCTA({ org }: { org: Org }) {
     return (
       <a
         href={`/pricing?org=${encodeURIComponent(org.slug)}`}
-        className="rounded bg-amber-600 px-4 py-2 text-sm font-medium text-white hover:bg-amber-500"
+        className="rounded bg-amber-800 px-4 py-2 text-sm font-medium text-white hover:bg-amber-700"
       >
         Complete payment
       </a>

canvas/src/components/AuditTrailPanel.tsx

@@ -8,11 +8,17 @@ import type { AuditEntry, AuditResponse } from "@/types/audit";
 
 type EventFilter = "all" | AuditEntry["event_type"];
 
+// Contrast note: text is rendered on near-black bg (bg-*-950/40). Every text
+// color below is chosen to pass WCAG 2.1 AA 4.5:1 on that background:
+//   blue-300   ( delegation ) ≈ 8.8:1
+//   violet-300 ( decision   ) ≈ 9.5:1
+//   yellow-200 ( gate       ) ≈ 11.5:1
+//   orange-300 ( hitl       ) ≈ 9.1:1
 const BADGE_COLORS: Record<AuditEntry["event_type"], { text: string; bg: string; border: string }> = {
-  delegation: { text: "text-accent",   bg: "bg-blue-950/40",   border: "border-blue-800/40" },
-  decision:   { text: "text-violet-400", bg: "bg-violet-950/40", border: "border-violet-800/40" },
-  gate:       { text: "text-yellow-400", bg: "bg-yellow-950/40", border: "border-yellow-800/40" },
-  hitl:       { text: "text-orange-400", bg: "bg-orange-950/40", border: "border-orange-800/40" },
+  delegation: { text: "text-blue-300",   bg: "bg-blue-950/40",   border: "border-blue-800/40" },
+  decision:   { text: "text-violet-300", bg: "bg-violet-950/40", border: "border-violet-800/40" },
+  gate:       { text: "text-yellow-200", bg: "bg-yellow-950/40", border: "border-yellow-800/40" },
+  hitl:       { text: "text-orange-300", bg: "bg-orange-950/40", border: "border-orange-800/40" },
 };
 
 const FILTERS: { id: EventFilter; label: string }[] = [
@@ -164,7 +170,10 @@ export function AuditTrailPanel({ workspaceId }: Props) {
 
       {/* Error banner */}
       {error && (
-        <div className="mx-4 mt-3 px-3 py-2 bg-red-950/30 border border-red-800/40 rounded text-xs text-bad shrink-0">
+        <div
+          role="alert"
+          className="mx-4 mt-3 px-3 py-2 bg-red-950/30 border border-red-800/40 rounded text-xs text-bad shrink-0"
+        >
           {error}
         </div>
       )}
@@ -242,7 +251,6 @@ export function AuditEntryRow({ entry, now }: AuditEntryRowProps) {
         {/* Event-type badge */}
         <span
           className={`shrink-0 text-[9px] font-semibold uppercase tracking-wider px-1.5 py-0.5 rounded border ${badge.text} ${badge.bg} ${badge.border}`}
-          aria-label={`Event type: ${entry.event_type}`}
         >
           {entry.event_type}
         </span>

canvas/src/components/BatchActionBar.tsx

@@ -100,8 +100,8 @@ export function BatchActionBar() {
       aria-label="Batch workspace actions"
       className="fixed bottom-6 left-1/2 -translate-x-1/2 z-[200] flex items-center gap-3 px-4 py-2.5 rounded-2xl bg-surface-sunken/95 border border-line/70 shadow-2xl shadow-black/50 backdrop-blur-md"
     >
-      {/* Selection count badge */}
-      <span className="text-[12px] font-semibold text-white bg-accent-strong/80 px-2.5 py-0.5 rounded-full tabular-nums">
+      {/* Selection count badge — bg-zinc-700 passes 7.2:1 on white text */}
+      <span className="text-[12px] font-semibold text-white bg-zinc-700 px-2.5 py-0.5 rounded-full tabular-nums">
         {count} selected
       </span>
 
@@ -112,7 +112,7 @@ export function BatchActionBar() {
         type="button"
         disabled={busy}
         onClick={() => setPending("restart")}
-        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-sky-300 bg-sky-900/30 hover:bg-sky-800/50 border border-sky-700/30 hover:border-sky-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-sky-500/70"
+        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-white bg-sky-900/30 hover:bg-sky-800/50 border border-sky-700/30 hover:border-sky-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-sky-500/70"
       >
         <span aria-hidden="true">↻</span>
         Restart All
@@ -122,7 +122,7 @@ export function BatchActionBar() {
         type="button"
         disabled={busy}
         onClick={() => setPending("pause")}
-        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-warm bg-amber-900/30 hover:bg-amber-800/50 border border-amber-700/30 hover:border-amber-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-amber-500/70"
+        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-white bg-amber-900/30 hover:bg-amber-800/50 border border-amber-700/30 hover:border-amber-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-amber-500/70"
       >
         <span aria-hidden="true">⏸</span>
         Pause All
@@ -132,7 +132,7 @@ export function BatchActionBar() {
         type="button"
         disabled={busy}
         onClick={() => setPending("delete")}
-        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-bad bg-red-900/30 hover:bg-red-800/50 border border-red-700/30 hover:border-red-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-500/70"
+        className="flex items-center gap-1.5 px-3 py-1.5 rounded-lg text-[12px] font-medium text-white bg-red-900/30 hover:bg-red-800/50 border border-red-700/30 hover:border-red-600/50 transition-colors disabled:opacity-50 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-500/70"
       >
         <span aria-hidden="true">✕</span>
         Delete All

canvas/src/components/ConfirmDialog.tsx

@@ -96,7 +96,7 @@ export function ConfirmDialog({
   // readable in both light and dark themes.
   const confirmColors =
     confirmVariant === "danger"
-      ? "bg-red-600 hover:bg-red-700 text-white"
+      ? "bg-red-700 hover:bg-red-600 text-white"
       : confirmVariant === "warning"
         ? "bg-amber-800 hover:bg-amber-700 text-white"
         : "bg-accent hover:bg-accent-strong text-white";

canvas/src/components/ContextMenu.tsx

@@ -318,7 +318,7 @@ export function ContextMenu() {
             aria-hidden="true"
             className={`w-1.5 h-1.5 rounded-full ${statusDotClass(contextMenu.nodeData.status)}`}
           />
-          <span className="text-[10px] text-ink-mid">{contextMenu.nodeData.status}</span>
+          <span className="text-[10px] text-ink">{contextMenu.nodeData.status}</span>
         </div>
       </div>
 

canvas/src/components/ConversationTraceModal.tsx

@@ -187,7 +187,7 @@ export function ConversationTraceModal({ open, workspaceId: _workspaceId, onClos
                                 isError
                                   ? "bg-red-950/50 text-bad"
                                   : isSend
-                                  ? "bg-cyan-950/50 text-cyan-400"
+                                  ? "bg-cyan-950 text-cyan-300"
                                   : isReceive
                                   ? "bg-blue-950/50 text-accent"
                                   : "bg-surface-card text-ink-mid"
@@ -251,7 +251,7 @@ export function ConversationTraceModal({ open, workspaceId: _workspaceId, onClos
 
                           {/* Error */}
                           {isError && entry.error_detail && (
-                            <div className="text-[10px] text-bad/80 mt-1 truncate">
+                            <div className="text-[10px] text-bad mt-1 truncate">
                               {entry.error_detail.slice(0, 200)}
                             </div>
                           )}
@@ -272,7 +272,7 @@ export function ConversationTraceModal({ open, workspaceId: _workspaceId, onClos
                           )}
                           {responseText && (
                             <div className="mt-1 bg-surface/60 border border-emerald-900/30 rounded-lg px-3 py-2 max-h-32 overflow-y-auto">
-                              <div className="text-[8px] text-good/60 uppercase mb-1">Response</div>
+                              <div className="text-[8px] text-good uppercase mb-1">Response</div>
                               <div className="text-[10px] text-ink-mid whitespace-pre-wrap break-words leading-relaxed">
                                 {responseText.slice(0, 2000)}
                                 {responseText.length > 2000 && (

canvas/src/components/DeleteCascadeConfirmDialog.tsx

@@ -126,8 +126,8 @@ export function DeleteCascadeConfirmDialog({
 
           {/* Cascade warning */}
           <div className="rounded border border-red-900/40 bg-red-950/20 px-3 py-2.5 mb-4">
-            <p className="text-[12px] text-bad/80 leading-relaxed">
-              Deleting will cascade — <strong className="text-red-200">all child workspaces and their data will be permanently removed.</strong> This cannot be undone.
+            <p className="text-[12px] text-red-300 leading-relaxed">
+              Deleting will cascade — <strong className="text-red-100">all child workspaces and their data will be permanently removed.</strong> This cannot be undone.
             </p>
           </div>
 
@@ -164,13 +164,13 @@ export function DeleteCascadeConfirmDialog({
             type="button"
             onClick={onConfirm}
             disabled={!checked}
-            // Hover goes DARKER, not lighter — bg-red-500 on white text
-            // drops contrast below AA vs bg-red-700. Same trap fixed in
-            // ConfirmDialog and ApprovalBanner. focus-visible ring matches.
+            // Hover goes DARKER, not lighter — bg-red-600 on white text
+            // drops contrast below AA. Same trap fixed in ConfirmDialog.
+            // focus-visible ring matches the canvas chrome.
             className={`px-3.5 py-1.5 text-[13px] rounded-lg transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-2 focus-visible:ring-offset-surface-sunken
               ${checked
-                ? "bg-red-600 hover:bg-red-700 text-white cursor-pointer"
-                : "bg-red-900/30 text-bad/40 cursor-not-allowed"
+                ? "bg-red-700 hover:bg-red-600 text-white cursor-pointer"
+                : "bg-red-900/30 text-red-400 cursor-not-allowed"
               }`}
           >
             Delete All

canvas/src/components/ErrorBoundary.tsx

@@ -51,7 +51,7 @@ export class ErrorBoundary extends React.Component<
   render() {
     if (this.state.hasError) {
       return (
-        <div className="fixed inset-0 flex items-center justify-center bg-surface z-50">
+        <div role="alert" aria-live="assertive" className="fixed inset-0 flex items-center justify-center bg-surface z-50">
           <div className="max-w-md rounded-2xl border border-red-500/30 bg-surface-sunken/90 px-8 py-8 text-center shadow-2xl shadow-black/40">
             <div className="mx-auto mb-4 flex h-14 w-14 items-center justify-center rounded-full bg-red-500/10 border border-red-500/30">
               <svg
@@ -76,7 +76,7 @@ export class ErrorBoundary extends React.Component<
             <p className="text-sm text-ink-mid mb-1">
               An unexpected error occurred while rendering the application.
             </p>
-            <p className="text-xs text-bad/80 mb-6 font-mono break-all">
+            <p className="text-xs text-bad mb-6 font-mono break-all">
               {this.state.error?.message ?? "Unknown error"}
             </p>
             <div className="flex items-center justify-center gap-3">

canvas/src/components/ExternalConnectModal.tsx

@@ -360,7 +360,7 @@ function SnippetBlock({
         <button
           type="button"
           onClick={onCopy}
-          className="text-xs px-2 py-1 rounded bg-accent-strong/80 hover:bg-accent text-white focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
+          className="text-xs px-2 py-1 rounded bg-accent text-white hover:bg-accent-strong transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
         >
           {copied ? "Copied!" : "Copy"}
         </button>

canvas/src/components/MissingKeysModal.tsx

@@ -344,7 +344,7 @@ function ProviderPickerModal({
   // wrapper's bounds instead of the viewport.
   if (typeof document === "undefined") return null;
 
-  const allSaved = entries.length > 0 && entries.every((e) => e.saved);
+  const allSaved = entries.every((e) => e.saved);
   const anySaving = entries.some((e) => e.saving);
   const runtimeLabel = runtime
     .replace(/[-_]/g, " ")
@@ -451,7 +451,7 @@ function ProviderPickerModal({
                     <button
                       onClick={() => handleSaveKey(index)}
                       disabled={!entry.value.trim() || entry.saving}
-                      className="px-3 py-1.5 bg-accent-strong hover:bg-accent text-[11px] rounded text-white disabled:opacity-30 transition-colors shrink-0"
+                      className="px-3 py-1.5 bg-accent-strong hover:bg-accent text-[11px] rounded text-white disabled:opacity-30 transition-colors shrink-0 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
                     >
                       {entry.saving ? "..." : "Save"}
                     </button>
@@ -492,7 +492,7 @@ function ProviderPickerModal({
                 !selectorValue.providerId ||
                 (showModelInput && model.trim() === "")
               }
-              className="px-3.5 py-1.5 text-[12px] bg-accent-strong hover:bg-accent text-white rounded-lg transition-colors disabled:opacity-40"
+              className="px-3.5 py-1.5 text-[12px] bg-accent-strong hover:bg-accent text-white rounded-lg transition-colors disabled:opacity-40 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
             >
               {allSaved ? "Deploy" : entries.length > 1 ? "Add Keys" : "Add Key"}
             </button>
@@ -616,7 +616,7 @@ function AllKeysModal({
   if (!open) return null;
   if (typeof document === "undefined") return null;
 
-  const allSaved = entries.length > 0 && entries.every((e) => e.saved);
+  const allSaved = entries.every((e) => e.saved);
   const anySaving = entries.some((e) => e.saving);
   const runtimeLabel = runtime
     .replace(/[-_]/g, " ")

canvas/src/components/ProviderModelSelector.tsx

@@ -420,7 +420,7 @@ export function ProviderModelSelector({
               spellCheck={false}
               autoComplete="off"
               data-testid="model-input"
-              className="w-full bg-surface-sunken border border-line rounded px-2 py-1.5 text-[11px] text-ink font-mono focus:outline-none focus:border-accent focus:ring-1 focus:ring-accent/20 transition-colors disabled:opacity-50"
+              className="w-full bg-surface-sunken border border-line rounded px-2 py-1.5 text-[11px] text-ink font-mono focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:border-accent transition-colors disabled:opacity-50"
             />
             <p className="text-[9px] text-ink-mid mt-1 leading-relaxed">
               {selected?.wildcard

canvas/src/components/ProvisioningTimeout.tsx

@@ -389,7 +389,7 @@ export function ProvisioningTimeout({
               <button
                 type="button"
                 onClick={handleCancelConfirm}
-                className="px-3.5 py-1.5 text-[12px] bg-red-600 hover:bg-red-500 text-white rounded-lg transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-400 focus-visible:ring-offset-1"
+                className="px-3.5 py-1.5 text-[12px] bg-red-800 hover:bg-red-700 text-white rounded-lg transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-400 focus-visible:ring-offset-1"
               >
                 Remove Workspace
               </button>

canvas/src/components/ThemeToggle.tsx

@@ -61,8 +61,12 @@ export function ThemeToggle({ className = "" }: { className?: string }) {
         return;
       }
       setTheme(OPTIONS[next].value);
-      // Move focus to the new button so arrow-key navigation is continuous
-      const btns = (e.currentTarget.closest("[role=radiogroup]") as HTMLElement)?.querySelectorAll<HTMLButtonElement>("[role=radio]");
+      // Move focus to the new button so arrow-key navigation is continuous.
+      // Query is already scoped to radiogroup so no child-combinator needed;
+      // avoids accidentally focusing unrelated [role=radio] elements
+      // elsewhere in the DOM (e.g. React Flow canvas nodes).
+      const radiogroup = e.currentTarget.closest("[role=radiogroup]") as HTMLElement | null;
+      const btns = radiogroup?.querySelectorAll<HTMLButtonElement>("[role=radio]");
       btns?.[next]?.focus();
     },
     []

canvas/src/components/WorkspaceNode.tsx

@@ -13,17 +13,20 @@ import { isExternalLikeRuntime } from "@/lib/externalRuntimes";
 
 /** Descendant count for the "N sub" badge — children are first-class nodes
  *  rendered as full cards inside this one via React Flow's native parentId,
- *  so we don't need to subscribe to the actual child list here. */
+ *  so we don't need to subscribe to the actual child list here.
+ *  Selecting `nodes` stably avoids a new selector reference on every store
+ *  update (React error #185 / Zustand + React 19 Object.is strictness). */
 function useDescendantCount(nodeId: string): number {
-  return useCanvasStore(
-    useCallback((s) => countDescendants(nodeId, s.nodes), [nodeId])
-  );
+  const nodes = useCanvasStore((s) => s.nodes);
+  return useMemo(() => countDescendants(nodeId, nodes), [nodeId, nodes]);
 }
 
+/** Boolean flag used to drive min-size and NodeResizer dimensions.
+ *  Selecting `nodes` stably avoids re-render loops (same issue as
+ *  useDescendantCount). */
 function useHasChildren(nodeId: string): boolean {
-  return useCanvasStore(
-    useCallback((s) => s.nodes.some((n) => n.data.parentId === nodeId), [nodeId])
-  );
+  const nodes = useCanvasStore((s) => s.nodes);
+  return useMemo(() => nodes.some((n) => n.data.parentId === nodeId), [nodes, nodeId]);
 }
 
 /** Eject/extract arrow icon — visually distinct from delete ✕ */

canvas/src/components/__tests__/ThemeToggle.test.tsx

@@ -24,8 +24,12 @@ vi.mock("@/lib/theme-provider", () => ({
   })),
 }));
 
+// Wrap cleanup in act() so any pending React state updates (e.g. from
+// keyDown handlers that call setTheme) flush before DOM unmount. Without
+// this, cleanup() can race against pending renders and cause INDEX_SIZE_ERR
+// when the handleKeyDown callback tries to query the DOM mid-teardown.
 afterEach(() => {
-  cleanup();
+  act(() => { cleanup(); });
   vi.clearAllMocks();
 });
 
@@ -146,7 +150,7 @@ describe("ThemeToggle — keyboard navigation (WCAG 2.1.1 / ARIA radiogroup)", (
     const radios = screen.getAllByRole("radio");
     // dark (index 2) is current; ArrowRight should wrap to light (index 0)
     act(() => { radios[2].focus(); });
-    fireEvent.keyDown(radios[2], { key: "ArrowRight" });
+    act(() => { fireEvent.keyDown(radios[2], { key: "ArrowRight" }); });
     expect(mockSetTheme).toHaveBeenCalledWith("light");
   });
 
@@ -160,7 +164,7 @@ describe("ThemeToggle — keyboard navigation (WCAG 2.1.1 / ARIA radiogroup)", (
     const radios = screen.getAllByRole("radio");
     // light (index 0) is current; ArrowLeft should go to dark (index 2)
     act(() => { radios[0].focus(); });
-    fireEvent.keyDown(radios[0], { key: "ArrowLeft" });
+    act(() => { fireEvent.keyDown(radios[0], { key: "ArrowLeft" }); });
     expect(mockSetTheme).toHaveBeenCalledWith("dark");
   });
 
@@ -174,7 +178,7 @@ describe("ThemeToggle — keyboard navigation (WCAG 2.1.1 / ARIA radiogroup)", (
     const radios = screen.getAllByRole("radio");
     // light (index 0) is current; ArrowDown should go to system (index 1)
     act(() => { radios[0].focus(); });
-    fireEvent.keyDown(radios[0], { key: "ArrowDown" });
+    act(() => { fireEvent.keyDown(radios[0], { key: "ArrowDown" }); });
     expect(mockSetTheme).toHaveBeenCalledWith("system");
   });
 
@@ -187,7 +191,7 @@ describe("ThemeToggle — keyboard navigation (WCAG 2.1.1 / ARIA radiogroup)", (
     render(<ThemeToggle />);
     const radios = screen.getAllByRole("radio");
     act(() => { radios[2].focus(); });
-    fireEvent.keyDown(radios[2], { key: "Home" });
+    act(() => { fireEvent.keyDown(radios[2], { key: "Home" }); });
     expect(mockSetTheme).toHaveBeenCalledWith("light");
   });
 
@@ -200,14 +204,14 @@ describe("ThemeToggle — keyboard navigation (WCAG 2.1.1 / ARIA radiogroup)", (
     render(<ThemeToggle />);
     const radios = screen.getAllByRole("radio");
     act(() => { radios[0].focus(); });
-    fireEvent.keyDown(radios[0], { key: "End" });
+    act(() => { fireEvent.keyDown(radios[0], { key: "End" }); });
     expect(mockSetTheme).toHaveBeenCalledWith("dark");
   });
 
   it("does nothing on unrelated keys", () => {
     render(<ThemeToggle />);
     const radios = screen.getAllByRole("radio");
-    fireEvent.keyDown(radios[0], { key: "Enter" });
+    act(() => { fireEvent.keyDown(radios[0], { key: "Enter" }); });
     expect(mockSetTheme).not.toHaveBeenCalled();
   });
 });

canvas/src/components/canvas/DropTargetBadge.tsx

@@ -24,16 +24,20 @@ import {
  */
 export function DropTargetBadge() {
   const dragOverNodeId = useCanvasStore((s) => s.dragOverNodeId);
-  const targetName = useCanvasStore((s) => {
-    if (!s.dragOverNodeId) return null;
-    const n = s.nodes.find((nn) => nn.id === s.dragOverNodeId);
+  // Select nodes stably first — deriving targetName and childCount inside
+  // the same selector creates a new return value on every store mutation
+  // even when neither has changed (React error #185 / Zustand Object.is).
+  const nodes = useCanvasStore((s) => s.nodes);
+  const targetName = (() => {
+    if (!dragOverNodeId) return null;
+    const n = nodes.find((nn) => nn.id === dragOverNodeId);
     return (n?.data as WorkspaceNodeData | undefined)?.name ?? null;
-  });
-  const childCount = useCanvasStore((s) =>
-    !s.dragOverNodeId
+  })();
+  const childCount = (() =>
+    !dragOverNodeId
       ? 0
-      : s.nodes.filter((n) => n.parentId === s.dragOverNodeId).length,
-  );
+      : nodes.filter((n) => n.parentId === dragOverNodeId).length
+  )();
   const { getInternalNode, flowToScreenPosition } = useReactFlow();
   if (!dragOverNodeId || !targetName) return null;
   const internal = getInternalNode(dragOverNodeId);
@@ -64,6 +68,7 @@ export function DropTargetBadge() {
       {ghostVisible && (
         <div
           data-testid="ghost-slot"
+          aria-hidden="true"
           className="pointer-events-none absolute z-40 rounded-lg border-2 border-dashed border-emerald-400/70 bg-emerald-500/10"
           style={{
             left: slotTL.x,
@@ -75,7 +80,9 @@ export function DropTargetBadge() {
       )}
       <div
         data-testid="drop-badge"
-        className="pointer-events-none absolute z-50 -translate-x-1/2 -translate-y-full rounded-md bg-emerald-500 px-2 py-0.5 text-[11px] font-medium text-white shadow-lg shadow-emerald-950/40"
+        role="status"
+        aria-label={`Drop target: ${targetName}`}
+        className="pointer-events-none absolute z-50 -translate-x-1/2 -translate-y-full rounded-md bg-emerald-700 px-2 py-0.5 text-[11px] font-medium text-white shadow-lg shadow-emerald-950/40"
         style={{ left: badge.x, top: badge.y - 6 }}
       >
         Drop into: {targetName}

canvas/src/components/canvas/__tests__/useOrgDeployState.test.ts

@@ -0,0 +1,311 @@
+/**
+ * Unit tests for buildDeployMap — the pure tree-traversal core of
+ * useOrgDeployState.
+ *
+ * What is tested here:
+ *   - Root / leaf identification via parent-chain walk
+ *   - isDeployingRoot: true when any descendant is "provisioning"
+ *   - isActivelyProvisioning: true only for the node itself in that state
+ *   - isLockedChild: true for non-root nodes in a deploying tree
+ *   - isLockedChild: also true for nodes in deletingIds (even if not deploying)
+ *   - descendantProvisioningCount: non-zero only on root nodes
+ *   - Performance contract: O(n) single-pass walk — tested by verifying
+ *     correctness across 50-node trees (n=50, all cases above)
+ *
+ * What is NOT tested here (hook integration — appropriate for E2E):
+ *   - The useMemo / Zustand subscription wiring
+ *   - React Flow integration (flowToScreenPosition, getInternalNode)
+ *
+ * Issue: #2071 (Canvas test gaps follow-up).
+ */
+import { describe, expect, it } from "vitest";
+import { buildDeployMap, type OrgDeployState } from "../useOrgDeployState";
+
+// ── Helpers ──────────────────────────────────────────────────────────────────
+
+type Projection = { id: string; parentId: string | null; status: string };
+
+function proj(
+  id: string,
+  parentId: string | null,
+  status: string,
+): Projection {
+  return { id, parentId, status };
+}
+
+/** Unchecked cast — test helpers aren't production code paths. */
+function m(
+  ps: Projection[],
+  deletingIds: string[] = [],
+): Map<string, OrgDeployState> {
+  return buildDeployMap(ps, new Set(deletingIds));
+}
+
+function s(
+  map: Map<string, OrgDeployState>,
+  id: string,
+): OrgDeployState {
+  const got = map.get(id);
+  if (!got) throw new Error(`no entry for id=${id}`);
+  return got;
+}
+
+// ── Empty / trivial ───────────────────────────────────────────────────────────
+
+describe("buildDeployMap — empty", () => {
+  it("returns empty map for empty projections", () => {
+    expect(m([]).size).toBe(0);
+  });
+});
+
+// ── Single node ─────────────────────────────────────────────────────────────
+
+describe("buildDeployMap — single node", () => {
+  it("isolated node is its own root and not deploying", () => {
+    const map = m([proj("a", null, "online")]);
+    expect(s(map, "a")).toEqual({
+      isActivelyProvisioning: false,
+      isDeployingRoot: false,
+      isLockedChild: false,
+      descendantProvisioningCount: 0,
+    });
+  });
+
+  it("isolated provisioning node is deploying root", () => {
+    const map = m([proj("a", null, "provisioning")]);
+    expect(s(map, "a")).toEqual({
+      isActivelyProvisioning: true,
+      isDeployingRoot: true,
+      isLockedChild: false,
+      descendantProvisioningCount: 1,
+    });
+  });
+});
+
+// ── Parent / child chains ─────────────────────────────────────────────────────
+
+describe("buildDeployMap — parent / child chains", () => {
+  it("root with online child: root is not deploying, child is not locked", () => {
+    // A ──► B
+    const map = m([
+      proj("A", null, "online"),
+      proj("B", "A", "online"),
+    ]);
+    expect(s(map, "A")).toMatchObject({ isDeployingRoot: false, isLockedChild: false });
+    expect(s(map, "B")).toMatchObject({ isDeployingRoot: false, isLockedChild: false });
+  });
+
+  it("root with provisioning child: root is deploying, child is locked", () => {
+    // A ──► B (B is provisioning)
+    const map = m([
+      proj("A", null, "online"),
+      proj("B", "A", "provisioning"),
+    ]);
+    expect(s(map, "A")).toMatchObject({ isDeployingRoot: true, descendantProvisioningCount: 1 });
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true, isActivelyProvisioning: true });
+  });
+
+  it("provisioning root with online child: root is deploying, child is locked", () => {
+    // A (provisioning) ──► B (online)
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "online"),
+    ]);
+    expect(s(map, "A")).toMatchObject({ isDeployingRoot: true, isActivelyProvisioning: true });
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true, isActivelyProvisioning: false });
+  });
+
+  it("grandchild inherits deploy lock through intermediate online node", () => {
+    // A ──► B ──► C  (A is provisioning)
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "online"),
+      proj("C", "B", "online"),
+    ]);
+    // B and C are both non-root descendants of the deploying root
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "C")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "A")).toMatchObject({ isDeployingRoot: true, descendantProvisioningCount: 1 });
+  });
+
+  it("deep chain: only the topmost node with a null parent counts as root", () => {
+    // A ──► B ──► C ──► D  (A is provisioning)
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "online"),
+      proj("C", "B", "online"),
+      proj("D", "C", "online"),
+    ]);
+    const roots = ["A", "B", "C", "D"].filter((id) => s(map, id).isDeployingRoot);
+    expect(roots).toEqual(["A"]);
+  });
+});
+
+// ── Sibling branching ─────────────────────────────────────────────────────────
+
+describe("buildDeployMap — sibling branching", () => {
+  it("parent with multiple children: deploying root propagates to all children", () => {
+    //         A (provisioning)
+    //        / \
+    //       B   C
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "online"),
+      proj("C", "A", "online"),
+    ]);
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "C")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "A")).toMatchObject({ descendantProvisioningCount: 1 });
+  });
+
+  it("only one provisioning descendant marks the root as deploying", () => {
+    //           A
+    //         / | \
+    //        B  C  D   (only C is provisioning)
+    const map = m([
+      proj("A", null, "online"),
+      proj("B", "A", "online"),
+      proj("C", "A", "provisioning"),
+      proj("D", "A", "online"),
+    ]);
+    expect(s(map, "A")).toMatchObject({ isDeployingRoot: true, descendantProvisioningCount: 1 });
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "C")).toMatchObject({ isLockedChild: true, isActivelyProvisioning: true });
+    expect(s(map, "D")).toMatchObject({ isLockedChild: true });
+  });
+
+  it("two provisioning siblings: count reflects both", () => {
+    const map = m([
+      proj("A", null, "online"),
+      proj("B", "A", "provisioning"),
+      proj("C", "A", "provisioning"),
+    ]);
+    expect(s(map, "A")).toMatchObject({ descendantProvisioningCount: 2 });
+    expect(s(map, "B")).toMatchObject({ isActivelyProvisioning: true });
+    expect(s(map, "C")).toMatchObject({ isActivelyProvisioning: true });
+  });
+});
+
+// ── Multiple disjoint trees ───────────────────────────────────────────────────
+
+describe("buildDeployMap — multiple disjoint trees", () => {
+  it("each tree has its own root; deploying nodes are independent", () => {
+    // Tree 1: X (provisioning) ──► Y
+    // Tree 2: P ──► Q  (no provisioning)
+    const map = m([
+      proj("X", null, "provisioning"),
+      proj("Y", "X", "online"),
+      proj("P", null, "online"),
+      proj("Q", "P", "online"),
+    ]);
+    expect(s(map, "X")).toMatchObject({ isDeployingRoot: true });
+    expect(s(map, "Y")).toMatchObject({ isLockedChild: true });
+    expect(s(map, "P")).toMatchObject({ isDeployingRoot: false, isLockedChild: false });
+    expect(s(map, "Q")).toMatchObject({ isDeployingRoot: false, isLockedChild: false });
+  });
+});
+
+// ── Deleting nodes ────────────────────────────────────────────────────────────
+
+describe("buildDeployMap — deletingIds", () => {
+  it("node in deletingIds is locked even if tree is not deploying", () => {
+    const map = m(
+      [
+        proj("A", null, "online"),
+        proj("B", "A", "online"),
+      ],
+      ["B"], // B is being deleted
+    );
+    expect(s(map, "A")).toMatchObject({ isLockedChild: false });
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true, isActivelyProvisioning: false });
+  });
+
+  it("node in deletingIds: isLockedChild is true regardless of provisioning", () => {
+    const map = m(
+      [
+        proj("A", null, "provisioning"),
+        proj("B", "A", "online"),
+      ],
+      ["B"],
+    );
+    // B is both a deploying-child AND a deleting node — either alone locks it
+    expect(s(map, "B")).toMatchObject({ isLockedChild: true });
+  });
+
+  it("empty deletingIds set has no effect", () => {
+    const map = m(
+      [
+        proj("A", null, "online"),
+        proj("B", "A", "online"),
+      ],
+      [],
+    );
+    expect(s(map, "B")).toMatchObject({ isLockedChild: false });
+  });
+});
+
+// ── descendantProvisioningCount ───────────────────────────────────────────────
+
+describe("buildDeployMap — descendantProvisioningCount", () => {
+  it("is 0 for non-root nodes", () => {
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "provisioning"),
+    ]);
+    expect(s(map, "B").descendantProvisioningCount).toBe(0);
+  });
+
+  it("includes the root's own status when provisioning", () => {
+    const map = m([
+      proj("A", null, "provisioning"),
+      proj("B", "A", "online"),
+    ]);
+    // A is both root and provisioning → count includes itself
+    expect(s(map, "A").descendantProvisioningCount).toBe(1);
+  });
+
+  it("accumulates all provisioning descendants (not just immediate children)", () => {
+    const map = m([
+      proj("A", null, "online"),
+      proj("B", "A", "online"),
+      proj("C", "B", "provisioning"),
+    ]);
+    expect(s(map, "A").descendantProvisioningCount).toBe(1);
+  });
+});
+
+// ── O(n) performance ─────────────────────────────────────────────────────────
+
+describe("buildDeployMap — O(n) performance contract", () => {
+  it("handles a 50-node three-level tree without incorrect node assignments", () => {
+    // Level 0: 1 root
+    // Level 1: 7 children
+    // Level 2: 42 leaves
+    // Total: 50 nodes
+    const projections: Projection[] = [];
+    projections.push(proj("root", null, "provisioning"));
+    for (let i = 0; i < 7; i++) {
+      projections.push(proj(`l1-${i}`, "root", "online"));
+    }
+    for (let i = 0; i < 42; i++) {
+      const parent = `l1-${Math.floor(i / 6)}`;
+      projections.push(proj(`l2-${i}`, parent, "online"));
+    }
+    const map = m(projections);
+
+    // Root is the only deploying node
+    expect(s(map, "root")).toMatchObject({
+      isDeployingRoot: true,
+      isLockedChild: false,
+      descendantProvisioningCount: 1,
+    });
+
+    // Every other node is a locked child
+    for (let i = 0; i < 7; i++) {
+      expect(s(map, `l1-${i}`)).toMatchObject({ isLockedChild: true, isDeployingRoot: false });
+    }
+    for (let i = 0; i < 42; i++) {
+      expect(s(map, `l2-${i}`)).toMatchObject({ isLockedChild: true, isDeployingRoot: false });
+    }
+  });
+});

canvas/src/components/canvas/useCanvasViewport.ts

@@ -1,6 +1,6 @@
 "use client";
 
-import { useCallback, useEffect, useRef } from "react";
+import { useCallback, useEffect, useMemo, useRef } from "react";
 import { useReactFlow } from "@xyflow/react";
 import { useCanvasStore } from "@/store/canvas";
 import { appendClass, removeClass } from "@/store/classNames";
@@ -153,10 +153,17 @@ export function useCanvasViewport() {
   // fit, the user has to manually pan + zoom to find what they just
   // created. Only fires when TRANSITIONING from some-provisioning to
   // zero-provisioning — not on every re-render.
-  const provisioningCount = useCanvasStore(
-    (s) => s.nodes.filter((n) => n.data.status === "provisioning").length,
+  //
+  // Selecting `nodes` stably (array reference) avoids the
+  // `.filter().length` anti-pattern which creates a new number on every
+  // store update and breaks the wasProvisioning/hasProvisioning
+  // transition detection (React error #185 / Zustand + React 19).
+  const nodes = useCanvasStore((s) => s.nodes);
+  const provisioningCount = useMemo(
+    () => nodes.filter((n) => n.data.status === "provisioning").length,
+    [nodes],
   );
-  const nodeCount = useCanvasStore((s) => s.nodes.length);
+  const nodeCount = nodes.length;
 
   useEffect(() => {
     const hasProvisioning = provisioningCount > 0;

canvas/src/components/mobile/MobileChat.tsx

@@ -5,22 +5,22 @@
 // that the desktop ChatTab uses, but with a slimmer surface: no
 // attachments, no A2A topology overlay, no conversation tracing.
 
-import { useEffect, useRef, useState } from "react";
+import { useEffect, useMemo, useRef, useState } from "react";
+import ReactMarkdown from "react-markdown";
+import remarkGfm from "remark-gfm";
 
-import { api } from "@/lib/api";
 import { useCanvasStore } from "@/store/canvas";
+import { type ChatAttachment, type ChatMessage, createMessage } from "@/components/tabs/chat/types";
+import {
+  useChatHistory,
+  useChatSend,
+  useChatSocket,
+} from "@/components/tabs/chat/hooks";
 
 import { toMobileAgent } from "./components";
 import { MOBILE_FONT_MONO, MOBILE_FONT_SANS, usePalette } from "./palette";
 import { Icons, StatusDot, TierChip } from "./primitives";
 
-interface ChatMessage {
-  id: string;
-  role: "user" | "agent" | "system";
-  text: string;
-  ts: string;
-}
-
 const formatStoredTimestamp = (iso: string): string => {
   const d = new Date(iso);
   if (isNaN(d.getTime())) return "";
@@ -29,15 +29,170 @@ const formatStoredTimestamp = (iso: string): string => {
 
 type SubTab = "my" | "a2a";
 
-interface A2AResponseShape {
-  result?: {
-    parts?: Array<{ kind?: string; text?: string }>;
-  };
-  error?: { message?: string };
-}
+function MarkdownBubble({
+  children,
+  dark,
+  accent,
+}: {
+  children: string;
+  dark: boolean;
+  accent: string;
+}) {
+  const codeBg = dark ? "rgba(255,255,255,0.08)" : "rgba(0,0,0,0.06)";
+  const codeBlockBg = dark ? "#1a1a1a" : "#f5f5f0";
+  const linkColor = accent;
+  const quoteBorder = dark ? "rgba(255,250,240,0.15)" : "rgba(40,30,20,0.15)";
 
-const formatTime = (date: Date) =>
-  date.toLocaleTimeString([], { hour: "numeric", minute: "2-digit" });
+  return (
+    <ReactMarkdown
+      remarkPlugins={[remarkGfm]}
+      components={{
+        p: ({ children }) => (
+          <div style={{ margin: "2px 0", lineHeight: "inherit" }}>{children}</div>
+        ),
+        a: ({ href, children }) => (
+          <a
+            href={href}
+            target="_blank"
+            rel="noopener noreferrer"
+            style={{ color: linkColor, textDecoration: "underline" }}
+          >
+            {children}
+          </a>
+        ),
+        pre: ({ children }) => (
+          <pre
+            style={{
+              background: codeBlockBg,
+              padding: "8px 10px",
+              borderRadius: 8,
+              overflow: "auto",
+              fontSize: 12,
+              lineHeight: 1.5,
+              fontFamily: MOBILE_FONT_MONO,
+              margin: "4px 0",
+            }}
+          >
+            {children}
+          </pre>
+        ),
+        code: ({ children, className }) => {
+          const isBlock = className != null && String(className).length > 0;
+          if (isBlock) {
+            return (
+              <code style={{ fontFamily: MOBILE_FONT_MONO, fontSize: 12 }}>
+                {children}
+              </code>
+            );
+          }
+          return (
+            <code
+              style={{
+                background: codeBg,
+                padding: "1px 4px",
+                borderRadius: 4,
+                fontSize: 13,
+                fontFamily: MOBILE_FONT_MONO,
+              }}
+            >
+              {children}
+            </code>
+          );
+        },
+        ul: ({ children }) => (
+          <ul style={{ margin: "4px 0", paddingLeft: 18, listStyle: "disc" }}>
+            {children}
+          </ul>
+        ),
+        ol: ({ children }) => (
+          <ol style={{ margin: "4px 0", paddingLeft: 18, listStyle: "decimal" }}>
+            {children}
+          </ol>
+        ),
+        li: ({ children }) => <li style={{ margin: "2px 0" }}>{children}</li>,
+        strong: ({ children }) => (
+          <strong style={{ fontWeight: 600 }}>{children}</strong>
+        ),
+        em: ({ children }) => <em style={{ fontStyle: "italic" }}>{children}</em>,
+        h1: ({ children }) => (
+          <div style={{ fontSize: 16, fontWeight: 700, margin: "4px 0" }}>{children}</div>
+        ),
+        h2: ({ children }) => (
+          <div style={{ fontSize: 15, fontWeight: 700, margin: "4px 0" }}>{children}</div>
+        ),
+        h3: ({ children }) => (
+          <div style={{ fontSize: 14, fontWeight: 700, margin: "4px 0" }}>{children}</div>
+        ),
+        h4: ({ children }) => (
+          <div style={{ fontSize: 14, fontWeight: 600, margin: "4px 0" }}>{children}</div>
+        ),
+        h5: ({ children }) => (
+          <div style={{ fontSize: 13, fontWeight: 600, margin: "4px 0" }}>{children}</div>
+        ),
+        h6: ({ children }) => (
+          <div style={{ fontSize: 13, fontWeight: 600, margin: "4px 0" }}>{children}</div>
+        ),
+        blockquote: ({ children }) => (
+          <blockquote
+            style={{
+              borderLeft: `2px solid ${quoteBorder}`,
+              margin: "4px 0",
+              paddingLeft: 8,
+              opacity: 0.85,
+            }}
+          >
+            {children}
+          </blockquote>
+        ),
+        hr: () => (
+          <hr
+            style={{
+              border: "none",
+              borderTop: `0.5px solid ${quoteBorder}`,
+              margin: "6px 0",
+            }}
+          />
+        ),
+        table: ({ children }) => (
+          <table
+            style={{
+              borderCollapse: "collapse",
+              fontSize: 13,
+              margin: "4px 0",
+              width: "100%",
+            }}
+          >
+            {children}
+          </table>
+        ),
+        thead: ({ children }) => <thead style={{ fontWeight: 600 }}>{children}</thead>,
+        th: ({ children }) => (
+          <th
+            style={{
+              border: `0.5px solid ${quoteBorder}`,
+              padding: "4px 6px",
+              textAlign: "left",
+            }}
+          >
+            {children}
+          </th>
+        ),
+        td: ({ children }) => (
+          <td
+            style={{
+              border: `0.5px solid ${quoteBorder}`,
+              padding: "4px 6px",
+            }}
+          >
+            {children}
+          </td>
+        ),
+      }}
+    >
+      {children}
+    </ReactMarkdown>
+  );
+}
 
 export function MobileChat({
   agentId,
@@ -49,34 +204,40 @@ export function MobileChat({
   onBack: () => void;
 }) {
   const p = usePalette(dark);
-  const node = useCanvasStore((s) => s.nodes.find((n) => n.id === agentId));
-  // Bootstrap from the canvas store's per-workspace message buffer so the
-  // user sees their prior thread on entry. The store is updated by the
-  // socket → ChatTab flows the desktop runs; on mobile we read from the
-  // same buffer to keep state coherent across viewports.
-  // NOTE: selector returns undefined (stable) — do NOT use ?? [] here,
-  // that creates a new [] reference on every store update when the key is
-  // absent, causing infinite re-render (React error #185).
-  const storedMessages = useCanvasStore((s) => s.agentMessages[agentId]);
-  const [messages, setMessages] = useState<ChatMessage[]>(() =>
-    (storedMessages ?? []).map((m) => ({
-      id: m.id,
-      role: "agent",
-      text: m.content,
-      ts: formatStoredTimestamp(m.timestamp),
-    })),
-  );
+  const nodes = useCanvasStore((s) => s.nodes);
+  const node = useMemo(() => nodes.find((n) => n.id === agentId), [nodes, agentId]);
   const [draft, setDraft] = useState("");
   const [tab, setTab] = useState<SubTab>("my");
-  const [sending, setSending] = useState(false);
-  const [error, setError] = useState<string | null>(null);
   const scrollRef = useRef<HTMLDivElement>(null);
-  // Synchronous re-entry guard. `setSending(true)` schedules a state
-  // update but doesn't flush before a second tap can fire send() — a ref
-  // mirrors the desktop ChatTab pattern (sendInFlightRef) and closes the
-  // double-send race a stale `sending` lets through.
-  const sendInFlightRef = useRef(false);
   const composerRef = useRef<HTMLTextAreaElement>(null);
+  const fileInputRef = useRef<HTMLInputElement>(null);
+  const [pendingFiles, setPendingFiles] = useState<File[]>([]);
+
+  const {
+    messages,
+    loading: historyLoading,
+    loadError: historyError,
+    loadInitial,
+    appendMessageDeduped,
+  } = useChatHistory(agentId);
+
+  const {
+    sending,
+    uploading,
+    sendMessage,
+    error: sendError,
+    clearError,
+    releaseSendGuards,
+  } = useChatSend(agentId, {
+    getHistoryMessages: () => messages,
+    onUserMessage: appendMessageDeduped,
+    onAgentMessage: appendMessageDeduped,
+  });
+
+  useChatSocket(agentId, {
+    onAgentMessage: appendMessageDeduped,
+    onSendComplete: releaseSendGuards,
+  });
 
   // Auto-grow the textarea: reset height to 'auto' so the scrollHeight
   // shrinks when the user deletes text, then size to scrollHeight up to
@@ -95,6 +256,20 @@ export function MobileChat({
     }
   }, [messages]);
 
+  // Consume any agent messages that arrived while history was loading.
+  const initialConsumeDoneRef = useRef(false);
+  useEffect(() => {
+    if (historyLoading || initialConsumeDoneRef.current) return;
+    initialConsumeDoneRef.current = true;
+    const consume = useCanvasStore.getState().consumeAgentMessages;
+    const msgs = consume(agentId);
+    for (const m of msgs) {
+      appendMessageDeduped(
+        createMessage("agent", m.content, m.attachments),
+      );
+    }
+  }, [historyLoading, agentId, appendMessageDeduped]);
+
   if (!node) {
     return (
       <div
@@ -116,58 +291,32 @@ export function MobileChat({
   const a = toMobileAgent(node);
   const reachable = a.status === "online" || a.status === "degraded";
 
+  const onFilesPicked = (fileList: FileList | null) => {
+    if (!fileList) return;
+    const picked = Array.from(fileList);
+    setPendingFiles((prev) => {
+      const keyed = new Set(prev.map((f) => `${f.name}:${f.size}`));
+      return [...prev, ...picked.filter((f) => !keyed.has(`${f.name}:${f.size}`))];
+    });
+    if (fileInputRef.current) fileInputRef.current.value = "";
+  };
+
+  const removePendingFile = (index: number) =>
+    setPendingFiles((prev) => prev.filter((_, i) => i !== index));
+
   const send = async () => {
     const text = draft.trim();
-    if (!text || sending || !reachable) return;
-    if (sendInFlightRef.current) return;
-    sendInFlightRef.current = true;
+    if ((!text && pendingFiles.length === 0) || sending || !reachable) return;
+    clearError();
     setDraft("");
-    setError(null);
-    setSending(true);
-    const myMsg: ChatMessage = {
-      id: crypto.randomUUID(),
-      role: "user",
-      text,
-      ts: formatTime(new Date()),
-    };
-    setMessages((m) => [...m, myMsg]);
-
-    try {
-      const res = await api.post<A2AResponseShape>(`/workspaces/${agentId}/a2a`, {
-        method: "message/send",
-        params: {
-          message: {
-            role: "user",
-            messageId: crypto.randomUUID(),
-            parts: [{ kind: "text", text }],
-          },
-        },
-      });
-      const reply =
-        res.result?.parts?.find((part) => part.kind === "text")?.text ?? "";
-      if (reply) {
-        setMessages((m) => [
-          ...m,
-          {
-            id: crypto.randomUUID(),
-            role: "agent",
-            text: reply,
-            ts: formatTime(new Date()),
-          },
-        ]);
-      } else if (res.error?.message) {
-        setError(res.error.message);
-      }
-    } catch (e) {
-      setError(e instanceof Error ? e.message : "Failed to send");
-    } finally {
-      setSending(false);
-      sendInFlightRef.current = false;
-    }
+    const files = pendingFiles;
+    setPendingFiles([]);
+    await sendMessage(text, files);
   };
 
   return (
     <div
+      data-testid="chat-panel"
       style={{
         height: "100%",
         display: "flex",
@@ -308,7 +457,42 @@ export function MobileChat({
             Agent Comms — peer-to-peer A2A traffic surfaces in the Comms tab.
           </div>
         )}
-        {tab === "my" && messages.length === 0 && (
+        {tab === "my" && historyLoading && (
+          <div style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
+            Loading chat history…
+          </div>
+        )}
+        {tab === "my" && !historyLoading && historyError && messages.length === 0 && (
+          <div
+            role="alert"
+            style={{
+              padding: "14px 4px",
+              textAlign: "center",
+              color: p.failed,
+              fontSize: 13,
+            }}
+          >
+            <div style={{ marginBottom: 8 }}>Could not load chat history.</div>
+            <button
+              type="button"
+              onClick={() => {
+                loadInitial();
+              }}
+              style={{
+                padding: "6px 14px",
+                borderRadius: 14,
+                border: `0.5px solid ${p.failed}`,
+                background: "transparent",
+                color: p.failed,
+                fontSize: 12,
+                cursor: "pointer",
+              }}
+            >
+              Retry
+            </button>
+          </div>
+        )}
+        {tab === "my" && !historyLoading && !historyError && messages.length === 0 && (
           <div style={{ padding: "20px 4px", textAlign: "center", color: p.text3, fontSize: 13 }}>
             Send a message to start chatting.
           </div>
@@ -337,7 +521,9 @@ export function MobileChat({
                     overflowWrap: "anywhere",
                   }}
                 >
-                  {m.text}
+                  <MarkdownBubble dark={dark} accent={p.accent}>
+                    {m.content}
+                  </MarkdownBubble>
                   <div
                     style={{
                       fontSize: 10,
@@ -346,13 +532,13 @@ export function MobileChat({
                       fontFamily: MOBILE_FONT_MONO,
                     }}
                   >
-                    {m.ts}
+                    {formatStoredTimestamp(m.timestamp)}
                   </div>
                 </div>
               </div>
             );
           })}
-        {error && (
+        {sendError && (
           <div
             role="alert"
             style={{
@@ -364,7 +550,7 @@ export function MobileChat({
               fontSize: 12,
             }}
           >
-            {error}
+            {sendError}
           </div>
         )}
       </div>
@@ -395,6 +581,60 @@ export function MobileChat({
           backdropFilter: "blur(14px)",
         }}
       >
+        {pendingFiles.length > 0 && (
+          <div
+            style={{
+              display: "flex",
+              flexWrap: "wrap",
+              gap: 6,
+              marginBottom: 8,
+              paddingLeft: 2,
+            }}
+          >
+            {pendingFiles.map((f, i) => (
+              <div
+                key={`${f.name}:${f.size}`}
+                style={{
+                  display: "flex",
+                  alignItems: "center",
+                  gap: 4,
+                  padding: "3px 8px",
+                  borderRadius: 10,
+                  background: dark ? "#2a2823" : "#ece9e0",
+                  fontSize: 12,
+                  color: p.text2,
+                  maxWidth: "100%",
+                }}
+              >
+                <span
+                  style={{
+                    overflow: "hidden",
+                    textOverflow: "ellipsis",
+                    whiteSpace: "nowrap",
+                  }}
+                >
+                  {f.name}
+                </span>
+                <button
+                  type="button"
+                  onClick={() => removePendingFile(i)}
+                  aria-label={`Remove ${f.name}`}
+                  style={{
+                    border: "none",
+                    background: "transparent",
+                    color: p.text3,
+                    cursor: "pointer",
+                    fontSize: 12,
+                    padding: 0,
+                    lineHeight: 1,
+                  }}
+                >
+                  ✕
+                </button>
+              </div>
+            ))}
+          </div>
+        )}
         <div
           style={{
             display: "flex",
@@ -406,21 +646,32 @@ export function MobileChat({
             padding: "6px 6px 6px 12px",
           }}
         >
+          <input
+            ref={fileInputRef}
+            type="file"
+            multiple
+            style={{ display: "none" }}
+            onChange={(e) => onFilesPicked(e.target.files)}
+            aria-hidden="true"
+          />
           <button
             type="button"
+            onClick={() => fileInputRef.current?.click()}
+            disabled={!reachable || sending || uploading}
             aria-label="Attach"
             style={{
               width: 32,
               height: 32,
               borderRadius: 999,
               border: "none",
-              cursor: "pointer",
+              cursor: reachable && !sending && !uploading ? "pointer" : "not-allowed",
               background: "transparent",
               color: p.text3,
               flexShrink: 0,
               display: "flex",
               alignItems: "center",
               justifyContent: "center",
+              opacity: !reachable || sending || uploading ? 0.4 : 1,
             }}
           >
             {Icons.attach({ size: 16 })}
@@ -466,28 +717,32 @@ export function MobileChat({
           <button
             type="button"
             onClick={send}
-            disabled={!draft.trim() || !reachable || sending}
+            disabled={(!draft.trim() && pendingFiles.length === 0) || !reachable || sending || uploading}
             aria-label="Send"
             style={{
               width: 36,
               height: 36,
               borderRadius: 999,
               border: "none",
-              cursor: draft.trim() && !sending ? "pointer" : "not-allowed",
+              cursor: (draft.trim() || pendingFiles.length > 0) && !sending && !uploading ? "pointer" : "not-allowed",
               flexShrink: 0,
               background:
-                draft.trim() && reachable && !sending
+                (draft.trim() || pendingFiles.length > 0) && reachable && !sending && !uploading
                   ? p.accent
                   : dark
                     ? "#2a2823"
                     : "#ece9e0",
-              color: draft.trim() && reachable && !sending ? "#fff" : p.text3,
+              color: (draft.trim() || pendingFiles.length > 0) && reachable && !sending && !uploading ? "#fff" : p.text3,
               display: "flex",
               alignItems: "center",
               justifyContent: "center",
             }}
           >
-            {Icons.send({ size: 16 })}
+            {uploading ? (
+              <span style={{ fontSize: 10, fontWeight: 600 }}>↑</span>
+            ) : (
+              Icons.send({ size: 16 })
+            )}
           </button>
         </div>
       </div>

canvas/src/components/mobile/MobileDetail.tsx

@@ -2,7 +2,7 @@
 
 // 03 · Agent detail — pills + tabbed content (Overview/Activity/Config/Memory).
 
-import { useEffect, useState } from "react";
+import { useEffect, useMemo, useState } from "react";
 
 import { api } from "@/lib/api";
 import { useCanvasStore } from "@/store/canvas";
@@ -32,7 +32,10 @@ export function MobileDetail({
   onChat: () => void;
 }) {
   const p = usePalette(dark);
-  const node = useCanvasStore((s) => s.nodes.find((n) => n.id === agentId));
+  // Selecting `nodes` stably avoids the `.find()` anti-pattern that
+  // creates a new return value on every store update (React error #185).
+  const nodes = useCanvasStore((s) => s.nodes);
+  const node = useMemo(() => nodes.find((n) => n.id === agentId), [nodes, agentId]);
   const [tab, setTab] = useState<TabId>("overview");
 
   if (!node) {
@@ -211,6 +214,7 @@ export function MobileDetail({
         <button
           type="button"
           onClick={onChat}
+          data-testid="mobile-chat-cta"
           style={{
             width: "100%",
             height: 52,

canvas/src/components/mobile/MobileSpawn.tsx

@@ -12,6 +12,7 @@ import { useEffect, useState } from "react";
 
 import { api } from "@/lib/api";
 import { type Template } from "@/lib/deploy-preflight";
+import { isSaaSTenant } from "@/lib/tenant";
 
 import { tierCode } from "./palette";
 import { MOBILE_FONT_MONO, MOBILE_FONT_SANS, type MobilePalette, usePalette } from "./palette";
@@ -26,6 +27,7 @@ const TIER_LABEL: Record<"T1" | "T2" | "T3" | "T4", string> = {
 
 export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => void }) {
   const p = usePalette(dark);
+  const isSaaS = isSaaSTenant();
   const [templates, setTemplates] = useState<Template[]>([]);
   const [loadingTemplates, setLoadingTemplates] = useState(true);
   const [tplId, setTplId] = useState<string | null>(null);
@@ -43,7 +45,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
         setTemplates(list);
         if (list.length > 0) {
           setTplId(list[0].id);
-          setTier(tierCode(list[0].tier));
+          setTier(isSaaS ? "T4" : tierCode(list[0].tier));
         }
       })
       .catch(() => {
@@ -55,7 +57,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
     return () => {
       cancelled = true;
     };
-  }, []);
+  }, [isSaaS]);
 
   const handleSpawn = async () => {
     if (busy || !tplId) return;
@@ -67,7 +69,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
       await api.post<{ id: string }>("/workspaces", {
         name: (name.trim() || chosen.name),
         template: chosen.id,
-        tier: Number(tier.slice(1)),
+        tier: isSaaS ? 4 : Number(tier.slice(1)),
         canvas: {
           x: Math.random() * 400 + 100,
           y: Math.random() * 300 + 100,
@@ -203,7 +205,7 @@ export function MobileSpawn({ dark, onClose }: { dark: boolean; onClose: () => v
             >
               {templates.map((t) => {
                 const on = tplId === t.id;
-                const tCode = tierCode(t.tier);
+                const tCode = isSaaS ? "T4" : tierCode(t.tier);
                 return (
                   <button
                     key={t.id}

canvas/src/components/mobile/__tests__/MobileChat.test.tsx

@@ -8,11 +8,19 @@
  * NOTE: No @testing-library/jest-dom — use DOM APIs.
  */
 import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
-import { cleanup, render } from "@testing-library/react";
+import { act, cleanup, render, waitFor } from "@testing-library/react";
 import React from "react";
 
 import { MobileChat } from "../MobileChat";
 
+// ─── Mock API ─────────────────────────────────────────────────────────────────
+// vi.mock without a factory auto-mocks the module. In tests, we configure
+// api.get / api.post directly (they are vi.fn() from the auto-mock).
+// Tests that need specific behaviour use mockResolvedValueOnce on the
+// auto-mocked functions.
+vi.mock("@/lib/api");
+import { api } from "@/lib/api";
+
 // ─── Mock store ───────────────────────────────────────────────────────────────
 
 const mockAgentId = "ws-chat-test";
@@ -28,12 +36,19 @@ const mockStoreState = {
     height?: number;
   }>,
   agentMessages: {} as Record<string, Array<{ id: string; content: string; timestamp: string }>>,
+  consumeAgentMessages: () => [],
 };
 
 vi.mock("@/store/canvas", () => ({
   useCanvasStore: Object.assign(
-    vi.fn((sel) => sel(mockStoreState)),
-    { getState: () => mockStoreState },
+    vi.fn((sel?: (state: typeof mockStoreState) => unknown) => {
+      if (sel) return sel(mockStoreState);
+      return mockStoreState;
+    }),
+    {
+      getState: () => mockStoreState,
+      subscribe: vi.fn(() => vi.fn()),
+    },
   ),
   summarizeWorkspaceCapabilities: vi.fn((data: Record<string, unknown>) => {
     const agentCard = data.agentCard as Record<string, unknown> | null;
@@ -54,16 +69,6 @@ vi.mock("@/store/canvas", () => ({
   }),
 }));
 
-// ─── Mock API ─────────────────────────────────────────────────────────────────
-
-const { mockApiPost } = vi.hoisted(() => ({
-  mockApiPost: vi.fn().mockResolvedValue({ result: { parts: [] } }),
-}));
-
-vi.mock("@/lib/api", () => ({
-  api: { post: mockApiPost },
-}));
-
 // ─── Fixtures ────────────────────────────────────────────────────────────────
 
 const onlineNode = {
@@ -150,7 +155,15 @@ beforeEach(() => {
   mockOnBack.mockClear();
   mockStoreState.nodes = [];
   mockStoreState.agentMessages = {};
-  mockApiPost.mockClear();
+  // 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");
+  getSpy.mockResolvedValue({ messages: [], reached_end: true });
+  postSpy.mockResolvedValue({ result: { parts: [] } });
+});
+
+afterEach(() => {
+  vi.restoreAllMocks();
 });
 
 afterEach(() => {
@@ -266,15 +279,26 @@ describe("MobileChat — empty state", () => {
     mockStoreState.nodes = [onlineNode];
   });
 
-  it('shows "Send a message to start chatting." when no messages', () => {
-    const { container } = renderChat(mockAgentId);
+  it('shows "Send a message to start chatting." when no messages', async () => {
+    // History fetch resolves immediately in tests (mockResolvedValue).
+    // act() flushes the microtask queue so the component reaches its
+    // post-load state before we assert.
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
     expect(container.textContent ?? "").toContain("Send a message to start chatting.");
   });
 
-  it("shows no messages when agentMessages[agentId] is absent (undefined)", () => {
+  it("shows no messages when agentMessages[agentId] is absent (undefined)", async () => {
     // Explicitly set to empty to simulate no stored messages
     mockStoreState.agentMessages = {};
-    const { container } = renderChat(mockAgentId);
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
     expect(container.textContent ?? "").toContain("Send a message to start chatting.");
   });
 });
@@ -321,3 +345,132 @@ describe("MobileChat — dark mode", () => {
     expect(container.querySelector('[aria-label="Back"]')).toBeTruthy();
   });
 });
+
+// ─── Chat history loading ────────────────────────────────────────────────────
+
+describe("MobileChat — chat history", () => {
+  beforeEach(() => {
+    mockStoreState.nodes = [onlineNode];
+  });
+
+  it("calls GET /workspaces/:id/chat-history on mount", async () => {
+    await act(async () => {
+      renderChat(mockAgentId);
+    });
+    expect(api.get).toHaveBeenCalledWith(
+      expect.stringContaining(`/workspaces/${mockAgentId}/chat-history`),
+    );
+  });
+
+  it("shows loading state while history is fetching", () => {
+    // Do NOT await — check the pre-resolve state.
+    const { container } = renderChat(mockAgentId);
+    expect(container.textContent ?? "").toContain("Loading chat history…");
+  });
+
+  it("shows empty state after history resolves with no messages", async () => {
+    // beforeEach already sets api.get to resolve with empty — no override needed.
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
+    expect(container.textContent ?? "").toContain("Send a message to start chatting.");
+  });
+
+  it("renders messages from history response", async () => {
+    vi.spyOn(api, "get").mockResolvedValueOnce({
+      messages: [
+        {
+          id: "msg-1",
+          role: "user",
+          content: "Hello agent",
+          timestamp: "2026-04-25T10:00:00Z",
+        },
+        {
+          id: "msg-2",
+          role: "agent",
+          content: "Hello back",
+          timestamp: "2026-04-25T10:00:01Z",
+        },
+      ],
+      reached_end: true,
+    });
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
+    expect(container.textContent ?? "").toContain("Hello agent");
+    expect(container.textContent ?? "").toContain("Hello back");
+  });
+
+  it("maps user role from API correctly", async () => {
+    vi.spyOn(api, "get").mockResolvedValueOnce({
+      messages: [
+        {
+          id: "msg-u",
+          role: "user",
+          content: "user message",
+          timestamp: "2026-04-25T10:00:00Z",
+        },
+      ],
+      reached_end: true,
+    });
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    // User messages render right-aligned. The text content check is sufficient
+    // to confirm the message appeared.
+    const { container } = renderResult!;
+    expect(container.textContent ?? "").toContain("user message");
+  });
+
+  it("shows error state when history fetch fails", async () => {
+    vi.spyOn(api, "get").mockRejectedValue(new Error("Network error"));
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
+    expect(container.textContent ?? "").toContain("Could not load chat history.");
+    expect(container.textContent ?? "").toContain("Retry");
+  });
+
+  it("Retry button re-fetches history after error", async () => {
+    // Make the initial mount call fail so the Retry button appears, then
+    // make the retry call succeed so we can verify the full flow.
+    const getSpy = vi.spyOn(api, "get");
+    getSpy
+      .mockRejectedValueOnce(new Error("Network error"))
+      .mockResolvedValueOnce({ messages: [], reached_end: true });
+
+    let renderResult: ReturnType<typeof renderChat>;
+    await act(async () => {
+      renderResult = renderChat(mockAgentId);
+    });
+    const { container } = renderResult!;
+
+    // Error state should be shown with Retry button.
+    expect(container.textContent ?? "").toContain("Could not load chat history.");
+    expect(container.textContent ?? "").toContain("Retry");
+
+    // Click Retry — the button's onClick fires api.get again.
+    // The second mockResolvedValueOnce makes it succeed.
+    const retryBtn = Array.from(container.querySelectorAll("button")).find(
+      (b) => b.textContent?.trim() === "Retry",
+    );
+    expect(retryBtn).toBeTruthy();
+    await act(async () => {
+      retryBtn?.click();
+    });
+
+    // waitFor polls until the retry resolves and component re-renders.
+    await waitFor(() => {
+      expect(container.textContent ?? "").toContain("Send a message to start chatting.");
+    });
+    // Initial call + retry = 2.
+    expect(getSpy).toHaveBeenCalledTimes(2);
+  });
+});

canvas/src/components/mobile/components.tsx

@@ -288,6 +288,7 @@ export function AgentCard({
   return (
     <button
       type="button"
+      data-testid="workspace-card"
       aria-label={`${agent.name}, status: ${agent.status}, tier ${agent.tier}${agent.remote ? ", remote" : ""}`}
       onClick={onClick}
       style={{

canvas/src/components/tabs/ActivityTab.tsx

@@ -307,7 +307,7 @@ function ActivityRow({
 
         {/* Error detail */}
         {isError && entry.error_detail && (
-          <div className="text-[9px] text-bad/80 mt-1 truncate">
+          <div className="text-[9px] text-bad mt-1 truncate">
             {entry.error_detail}
           </div>
         )}
@@ -358,10 +358,10 @@ function A2AErrorPreview({ label, raw }: { label: string; raw: string }) {
   const hint = inferA2AErrorHint(detail);
   return (
     <div>
-      <div className="text-[8px] text-bad/80 uppercase tracking-wider mb-1">{label} — delivery failed</div>
+      <div className="text-[8px] text-bad uppercase tracking-wider mb-1">{label} — delivery failed</div>
       <div className="text-[10px] text-bad bg-red-950/30 border border-red-800/40 rounded p-2 space-y-1.5">
         <div className="font-mono whitespace-pre-wrap break-words max-h-32 overflow-y-auto">{detail}</div>
-        <div className="text-[9px] text-bad/70 leading-relaxed border-t border-red-800/30 pt-1.5">{hint}</div>
+        <div className="text-[9px] text-bad leading-relaxed border-t border-red-800/30 pt-1.5">{hint}</div>
       </div>
     </div>
   );

canvas/src/components/tabs/BudgetSection.tsx

@@ -243,7 +243,7 @@ export function BudgetSection({ workspaceId }: Props) {
           onClick={handleSave}
           disabled={saving}
           data-testid="budget-save-btn"
-          className="px-4 py-1.5 bg-accent-strong hover:bg-accent active:bg-accent-strong rounded-lg text-xs font-medium text-white disabled:opacity-50 transition-colors"
+          className="px-4 py-1.5 bg-accent-strong hover:bg-accent active:bg-accent-strong rounded-lg text-xs font-medium text-white disabled:opacity-50 transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
         >
           {saving ? "Saving…" : "Save"}
         </button>

canvas/src/components/tabs/ChannelsTab.tsx

@@ -255,7 +255,7 @@ export function ChannelsTab({ workspaceId }: Props) {
         </h3>
         <button
           onClick={() => setShowForm(!showForm)}
-          className="text-[10px] px-2.5 py-1 rounded bg-accent-strong/20 text-accent hover:bg-accent-strong/30 transition"
+          className="text-[10px] px-2.5 py-1 rounded bg-accent-strong/20 text-accent hover:bg-accent-strong/30 transition focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
         >
           {showForm ? "Cancel" : "+ Connect"}
         </button>
@@ -308,7 +308,7 @@ export function ChannelsTab({ workspaceId }: Props) {
                             <button
                               onClick={handleDiscover}
                               disabled={discovering || !formValues["bot_token"]}
-                              className="text-[10px] px-2 py-0.5 rounded bg-accent-strong/20 text-accent hover:bg-accent-strong/30 transition disabled:opacity-40"
+                              className="text-[10px] px-2 py-0.5 rounded bg-accent-strong/20 text-accent hover:bg-accent-strong/30 transition disabled:opacity-40 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
                             >
                               {discovering ? "Detecting..." : "Detect Chats"}
                             </button>

canvas/src/components/tabs/ConfigTab.tsx

@@ -176,7 +176,7 @@ export function deriveProvidersFromModels(models: ModelSpec[]): string[] {
 // exactly the point of the platform adaptor. The deep `~/.hermes/
 // config.yaml` on the container is a separate runtime-internal file,
 // not this one.
-const RUNTIMES_WITH_OWN_CONFIG = new Set<string>(["external", "kimi", "kimi-cli"]);
+const RUNTIMES_WITH_OWN_CONFIG = new Set<string>(["external", "kimi", "kimi-cli", "openclaw"]);
 
 const FALLBACK_RUNTIME_OPTIONS: RuntimeOption[] = [
   { value: "", label: "LangGraph (default)", models: [], providers: [] },

canvas/src/components/tabs/DetailsTab.tsx

@@ -325,10 +325,10 @@ export function DetailsTab({ workspaceId, data }: Props) {
               <button
                 type="button"
                 onClick={handleDelete}
-                // hover:bg-red-500 LIGHTER on white text drops AA;
-                // flipped to bg-red-700 + focus-visible danger ring,
-                // matching the ConfirmDialog/DeleteCascade pattern.
-                className="px-3 py-1 bg-red-600 hover:bg-red-700 text-xs rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface"
+                // Red-600 on white text = 3.9:1 (WCAG AA FAIL).
+                // Red-700 = 4.6:1 (PASS). Hover goes DARKER (red-600)
+                // to signal press. Same pattern as ConfirmDialog/DeleteCascade.
+                className="px-3 py-1 bg-red-700 hover:bg-red-600 text-xs rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface"
               >
                 Confirm Delete
               </button>

canvas/src/components/tabs/ExternalConnectionSection.tsx

@@ -131,7 +131,7 @@ export function ExternalConnectionSection({ workspaceId }: Props) {
               <button
                 type="button"
                 onClick={doRotate}
-                className="px-3 py-1.5 bg-red-700 hover:bg-red-600 text-xs rounded text-white focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-500 focus-visible:ring-offset-1"
+                className="px-3 py-1.5 bg-red-800 hover:bg-red-700 text-xs rounded text-white focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-500 focus-visible:ring-offset-1"
               >
                 Rotate
               </button>

canvas/src/components/tabs/FilesTab.tsx

@@ -226,7 +226,7 @@ function PlatformOwnedFilesTab({ workspaceId }: { workspaceId: string }) {
         <div role="alertdialog" aria-labelledby="files-delete-all-msg" className="mx-3 mt-2 px-3 py-2 bg-red-950/30 border border-red-800/40 rounded space-y-1.5">
           <p id="files-delete-all-msg" className="text-xs text-bad">Delete all {files.filter((f) => !f.dir).length} files? This cannot be undone.</p>
           <div className="flex gap-2">
-            <button type="button" onClick={() => { handleDeleteAll(); setShowDeleteAll(false); }} className="px-2 py-0.5 bg-red-600 hover:bg-red-700 text-[10px] rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Delete All</button>
+            <button type="button" onClick={() => { handleDeleteAll(); setShowDeleteAll(false); }} className="px-2 py-0.5 bg-red-700 hover:bg-red-600 text-[10px] rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Delete All</button>
             <button type="button" onClick={() => setShowDeleteAll(false)} className="px-2 py-0.5 bg-surface-card hover:bg-surface-elevated hover:text-ink text-[10px] rounded text-ink-mid transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-accent/40 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Cancel</button>
           </div>
         </div>
@@ -240,7 +240,7 @@ function PlatformOwnedFilesTab({ workspaceId }: { workspaceId: string }) {
         <div role="alertdialog" aria-labelledby="files-delete-one-msg" className="mx-3 mt-2 px-3 py-2 bg-amber-950/30 border border-amber-800/40 rounded space-y-1.5">
           <p id="files-delete-one-msg" className="text-xs text-warm">Delete <span className="font-mono">{confirmDelete}</span>{files.find((f) => f.path === confirmDelete && f.dir) ? " and all its contents" : ""}?</p>
           <div className="flex gap-2">
-            <button type="button" onClick={confirmDeleteFile} className="px-2 py-0.5 bg-red-600 hover:bg-red-700 text-[10px] rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Delete</button>
+            <button type="button" onClick={confirmDeleteFile} className="px-2 py-0.5 bg-red-700 hover:bg-red-600 text-[10px] rounded text-white transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-red-500/60 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Delete</button>
             <button type="button" onClick={() => setConfirmDelete(null)} className="px-2 py-0.5 bg-surface-card hover:bg-surface-elevated hover:text-ink text-[10px] rounded text-ink-mid transition-colors focus:outline-none focus-visible:ring-2 focus-visible:ring-accent/40 focus-visible:ring-offset-1 focus-visible:ring-offset-surface">Cancel</button>
           </div>
         </div>

canvas/src/components/tabs/FilesTab/FilesToolbar.tsx

@@ -32,7 +32,7 @@ export function FilesToolbar({
           value={root}
           onChange={(e) => setRoot(e.target.value)}
           aria-label="File root directory"
-          className="text-[10px] bg-surface-card text-ink-mid border border-line rounded px-1.5 py-0.5 outline-none"
+          className="text-[10px] bg-surface-card text-ink-mid border border-line rounded px-1.5 py-0.5 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1"
         >
           <option value="/configs">/configs</option>
           <option value="/home">/home</option>

canvas/src/components/tabs/FilesTab/__tests__/FilesTab.test.tsx

@@ -1,217 +1,181 @@
 // @vitest-environment jsdom
 /**
- * FilesTab: NotAvailablePanel + FilesToolbar coverage.
+ * Tests for the main FilesTab / PlatformOwnedFilesTab component.
  *
- * NotAvailablePanel: pure presentational component — renders a "feature not
- * available" placeholder for external-runtime workspaces.
- * FilesToolbar: pure props-driven component — directory selector, file count,
- * action buttons (New, Upload, Export, Clear, Refresh) with correct aria-labels.
+ * Covers: NotAvailablePanel (external runtime), loading/empty/error states,
+ * FilesToolbar actions, and the /configs-only upload guard.
  *
- * No @testing-library/jest-dom import — use textContent / className /
- * getAttribute checks to avoid "expect is not defined" errors.
+ * No @testing-library/jest-dom — use textContent / className / getAttribute.
  */
 import { afterEach, describe, expect, it, vi } from "vitest";
-import { cleanup, render, screen } from "@testing-library/react";
+import { cleanup, fireEvent, render, screen, waitFor } from "@testing-library/react";
 import React from "react";
 
-import { FilesToolbar } from "../FilesToolbar";
-import { NotAvailablePanel } from "../NotAvailablePanel";
+import { FilesTab } from "../../FilesTab.tsx";
+import { FilesToolbar } from "../FilesToolbar.tsx";
+import type { FileEntry } from "../../FilesTab/tree";
 
-// ─── afterEach ─────────────────────────────────────────────────────────────────
+// ─── Mock ──────────────────────────────────────────────────────────────────
+
+const _mockGet = vi.hoisted(() => vi.fn<() => Promise<unknown>>());
+vi.mock("@/lib/api", () => ({
+  api: { get: _mockGet, put: vi.fn(), del: vi.fn() },
+}));
 
 afterEach(() => {
   cleanup();
-  vi.restoreAllMocks();
+  _mockGet.mockReset();
 });
 
-// ─── NotAvailablePanel ─────────────────────────────────────────────────────────
+// ─── Helpers ───────────────────────────────────────────────────────────────
 
-describe("NotAvailablePanel", () => {
-  it("renders heading 'Files not available'", () => {
-    const { container } = render(<NotAvailablePanel runtime="external" />);
-    expect(container.textContent).toContain("Files not available");
-  });
+const emptyFileList: FileEntry[] = [];
 
-  it("renders the runtime name in monospace", () => {
-    const { container } = render(<NotAvailablePanel runtime="external" />);
-    expect(container.textContent).toContain("external");
-    const spans = container.querySelectorAll("span");
-    const monoSpans = Array.from(spans).filter(
-      (s) => s.className && s.className.includes("font-mono"),
-    );
-    expect(monoSpans.length).toBeGreaterThan(0);
-  });
+/** Render FilesTab with a non-external runtime (triggers PlatformOwnedFilesTab). */
+function renderPlatformTab(extraProps: Partial<React.ComponentProps<typeof FilesTab>> = {}) {
+  return render(
+    <FilesTab
+      workspaceId="ws-1"
+      data={{ id: "ws-1", name: "Test", runtime: "claude-code", status: "online", tier: 0, skills: [], created_at: "" }}
+      {...extraProps}
+    />,
+  );
+}
 
-  it("renders a Chat tab hint in description", () => {
-    const { container } = render(<NotAvailablePanel runtime="remote-agent" />);
-    expect(container.textContent).toContain("Chat tab");
-  });
+/** Render FilesToolbar directly with stub handlers. */
+function renderToolbar(extraProps: Partial<React.ComponentProps<typeof FilesToolbar>> = {}) {
+  return render(
+    <FilesToolbar
+      root="/configs"
+      setRoot={vi.fn()}
+      fileCount={0}
+      onNewFile={vi.fn()}
+      onUpload={vi.fn()}
+      onDownloadAll={vi.fn()}
+      onClearAll={vi.fn()}
+      onRefresh={vi.fn()}
+      {...extraProps}
+    />
+  );
+}
 
-  it("SVG icon has aria-hidden=true", () => {
-    const { container } = render(<NotAvailablePanel runtime="external" />);
-    const svg = container.querySelector("svg");
-    expect(svg?.getAttribute("aria-hidden")).toBe("true");
-  });
+// ─── NotAvailablePanel ──────────────────────────────────────────────────────
 
-  it("renders without crashing for any runtime string", () => {
-    const { container } = render(<NotAvailablePanel runtime="unknown-runtime" />);
-    expect(container.textContent).toContain("unknown-runtime");
-  });
-
-  it("applies the correct layout classes to root div", () => {
-    const { container } = render(<NotAvailablePanel runtime="external" />);
-    const root = container.firstElementChild as HTMLElement;
-    expect(root.className).toContain("flex");
-    expect(root.className).toContain("flex-col");
-    expect(root.className).toContain("items-center");
-  });
-});
-
-// ─── FilesToolbar ───────────────────────────────────────────────────────────────
-
-describe("FilesToolbar", () => {
-  const noop = vi.fn();
-
-  function renderToolbar(props: Partial<React.ComponentProps<typeof FilesToolbar>> = {}) {
-    return render(
-      <FilesToolbar
-        root="/configs"
-        setRoot={noop}
-        fileCount={0}
-        onNewFile={noop}
-        onUpload={noop}
-        onDownloadAll={noop}
-        onClearAll={noop}
-        onRefresh={noop}
-        {...props}
+describe("FilesTab — NotAvailablePanel", () => {
+  it("renders NotAvailablePanel when runtime is external", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    render(
+      <FilesTab
+        workspaceId="ws-1"
+        data={{ id: "ws-1", name: "Test", runtime: "external", status: "online", tier: 0, skills: [], created_at: "" }}
       />,
     );
-  }
-
-  it("renders the directory selector with correct aria-label", () => {
-    const { container } = renderToolbar();
-    const select = container.querySelector("select");
-    expect(select?.getAttribute("aria-label")).toBe("File root directory");
+    expect(screen.getByText(/Files not available/i)).toBeTruthy();
   });
 
-  it("directory selector has all four options", () => {
-    const { container } = renderToolbar();
-    const select = container.querySelector("select") as HTMLSelectElement;
-    const options = Array.from(select?.options ?? []);
-    const values = options.map((o) => o.value);
-    expect(values).toContain("/configs");
-    expect(values).toContain("/home");
-    expect(values).toContain("/workspace");
-    expect(values).toContain("/plugins");
-  });
-
-  it("calls setRoot when directory changes", () => {
-    const setRoot = vi.fn();
-    const { container } = renderToolbar({ setRoot });
-    const select = container.querySelector("select") as HTMLSelectElement;
-    select.value = "/home";
-    select.dispatchEvent(new Event("change", { bubbles: true }));
-    expect(setRoot).toHaveBeenCalledWith("/home");
-  });
-
-  it("displays the file count", () => {
-    const { container } = renderToolbar({ fileCount: 42 });
-    expect(container.textContent).toContain("42 files");
-  });
-
-  it("shows New + Upload + Clear buttons for /configs", () => {
-    const { container } = renderToolbar({ root: "/configs" });
-    const texts = Array.from(container.querySelectorAll("button")).map(
-      (b) => b.textContent?.trim(),
+  it("renders the runtime name in NotAvailablePanel", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    render(
+      <FilesTab
+        workspaceId="ws-1"
+        data={{ id: "ws-1", name: "Test", runtime: "external", status: "online", tier: 0, skills: [], created_at: "" }}
+      />,
     );
-    expect(texts).toContain("+ New");
-    expect(texts).toContain("Upload");
-    expect(texts).toContain("Clear");
-    expect(texts).toContain("Export");
-    expect(texts).toContain("↻");
+    expect(screen.getByText(/external/i)).toBeTruthy();
   });
 
-  it("hides New + Upload + Clear for /workspace", () => {
-    const { container } = renderToolbar({ root: "/workspace" });
-    const texts = Array.from(container.querySelectorAll("button")).map(
-      (b) => b.textContent?.trim(),
+  it("does NOT call api.get when runtime is external", async () => {
+    render(
+      <FilesTab
+        workspaceId="ws-1"
+        data={{ id: "ws-1", name: "Test", runtime: "external", status: "online", tier: 0, skills: [], created_at: "" }}
+      />,
     );
-    expect(texts).not.toContain("+ New");
-    expect(texts).not.toContain("Upload");
-    expect(texts).not.toContain("Clear");
-    expect(texts).toContain("Export");
+    expect(_mockGet).not.toHaveBeenCalled();
   });
+});
 
-  it("hides New + Upload + Clear for /home", () => {
-    const { container } = renderToolbar({ root: "/home" });
-    const texts = Array.from(container.querySelectorAll("button")).map(
-      (b) => b.textContent?.trim(),
+// ─── Loading / Empty / Error states ────────────────────────────────────────
+
+describe("FilesTab — states", () => {
+  it("shows loading text while fetching files", () => {
+    _mockGet.mockImplementation(
+      () => new Promise<unknown>(() => {}) as unknown as Promise<unknown>,
     );
-    expect(texts).not.toContain("+ New");
-    expect(texts).not.toContain("Upload");
-    expect(texts).not.toContain("Clear");
+    renderPlatformTab();
+    expect(screen.getByText("Loading files...")).toBeTruthy();
   });
 
-  it("hides New + Upload + Clear for /plugins", () => {
-    const { container } = renderToolbar({ root: "/plugins" });
-    const texts = Array.from(container.querySelectorAll("button")).map(
-      (b) => b.textContent?.trim(),
-    );
-    expect(texts).not.toContain("+ New");
-    expect(texts).not.toContain("Upload");
-    expect(texts).not.toContain("Clear");
+  it("shows 'No config files yet' when root is /configs and no files", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => {
+      expect(screen.getByText(/No config files yet/i)).toBeTruthy();
+    });
   });
 
-  it("New button has correct aria-label", () => {
-    const { container } = renderToolbar({ root: "/configs" });
-    const newBtn = container.querySelector('button[aria-label="Create new file"]');
-    expect(newBtn?.textContent?.trim()).toBe("+ New");
+  it("fetches from the correct endpoint", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => {
+      expect(_mockGet).toHaveBeenCalledWith(expect.stringContaining("/workspaces/ws-1/files"));
+    });
   });
 
-  it("Export button has correct aria-label", () => {
-    const { container } = renderToolbar();
-    const exportBtn = container.querySelector('button[aria-label="Download all files"]');
-    expect(exportBtn?.textContent?.trim()).toBe("Export");
+  it("shows file count from toolbar when files exist", async () => {
+    _mockGet.mockResolvedValue([
+      { path: "configs/a.yaml", size: 10, dir: false },
+      { path: "configs/b.yaml", size: 20, dir: false },
+    ]);
+    renderPlatformTab();
+    await waitFor(() => {
+      expect(screen.getByText("2 files")).toBeTruthy();
+    });
+  });
+});
+
+// ─── FilesToolbar ──────────────────────────────────────────────────────────
+
+describe("FilesTab — FilesToolbar", () => {
+  it("shows Refresh button", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => {
+      expect(screen.getByLabelText("Refresh file list")).toBeTruthy();
+    });
   });
 
-  it("Clear button has correct aria-label", () => {
-    const { container } = renderToolbar({ root: "/configs" });
-    const clearBtn = container.querySelector('button[aria-label="Delete all files"]');
-    expect(clearBtn?.textContent?.trim()).toBe("Clear");
+  it("shows root directory selector", async () => {
+    _mockGet.mockResolvedValueOnce(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => {
+      expect(screen.getByRole("combobox")).toBeTruthy();
+    });
   });
 
-  it("Refresh button has correct aria-label", () => {
-    const { container } = renderToolbar();
-    const refreshBtn = container.querySelector('button[aria-label="Refresh file list"]');
-    expect(refreshBtn?.textContent?.trim()).toBe("↻");
+  it("Refresh button triggers a reload", async () => {
+    // Use persistent mock — loadFiles fires on mount AND on Refresh click.
+    _mockGet.mockResolvedValue(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => screen.getByLabelText("Refresh file list"));
+    const before = _mockGet.mock.calls.length;
+    fireEvent.click(screen.getByLabelText("Refresh file list"));
+    await waitFor(() => {
+      expect(_mockGet.mock.calls.length).toBeGreaterThan(before);
+    });
   });
+});
 
-  it("calls onNewFile when New button is clicked", () => {
-    const onNewFile = vi.fn();
-    const { container } = renderToolbar({ root: "/configs", onNewFile });
-    container.querySelector('button[aria-label="Create new file"]')!.click();
-    expect(onNewFile).toHaveBeenCalledTimes(1);
-  });
+// ─── Upload guard ──────────────────────────────────────────────────────────
 
-  it("calls onDownloadAll when Export button is clicked", () => {
-    const onDownloadAll = vi.fn();
-    const { container } = renderToolbar({ onDownloadAll });
-    container.querySelector('button[aria-label="Download all files"]')!.click();
-    expect(onDownloadAll).toHaveBeenCalledTimes(1);
-  });
+describe("FilesTab — upload guard", () => {
+  it("no error alert on dragover when root is /configs (default)", async () => {
+    _mockGet.mockResolvedValue(emptyFileList);
+    renderPlatformTab();
+    await waitFor(() => screen.getByText(/No config files yet/i));
 
-  it("calls onClearAll when Clear button is clicked", () => {
-    const onClearAll = vi.fn();
-    const { container } = renderToolbar({ root: "/configs", onClearAll });
-    container.querySelector('button[aria-label="Delete all files"]')!.click();
-    expect(onClearAll).toHaveBeenCalledTimes(1);
-  });
-
-  it("calls onRefresh when Refresh button is clicked", () => {
-    const onRefresh = vi.fn();
-    const { container } = renderToolbar({ onRefresh });
-    container.querySelector('button[aria-label="Refresh file list"]')!.click();
-    expect(onRefresh).toHaveBeenCalledTimes(1);
+    // No alert should be present
+    expect(screen.queryByRole("alert")).toBeNull();
   });
 
   it("applies focus-visible ring to all interactive buttons", () => {

canvas/src/components/tabs/FilesTab/__tests__/tree.test.ts

@@ -0,0 +1,218 @@
+// @vitest-environment jsdom
+/**
+ * Tests for tree.ts — buildTree and getIcon pure functions.
+ */
+import { describe, expect, it } from "vitest";
+import type { FileEntry } from "../tree";
+import { buildTree, getIcon } from "../tree";
+
+// ─── getIcon ─────────────────────────────────────────────────────────────────
+
+describe("getIcon", () => {
+  it("returns folder emoji for directories", () => {
+    expect(getIcon("/configs", true)).toBe("📁");
+  });
+
+  it("returns correct emoji for .md", () => {
+    expect(getIcon("readme.md", false)).toBe("📄");
+  });
+
+  it("returns correct emoji for .yaml", () => {
+    expect(getIcon("config.yaml", false)).toBe("⚙");
+  });
+
+  it("returns correct emoji for .yml", () => {
+    expect(getIcon("config.yml", false)).toBe("⚙");
+  });
+
+  it("returns correct emoji for .py", () => {
+    expect(getIcon("script.py", false)).toBe("🐍");
+  });
+
+  it("returns correct emoji for .ts", () => {
+    expect(getIcon("index.ts", false)).toBe("💠");
+  });
+
+  it("returns correct emoji for .tsx", () => {
+    expect(getIcon("App.tsx", false)).toBe("💠");
+  });
+
+  it("returns correct emoji for .js", () => {
+    expect(getIcon("index.js", false)).toBe("📜");
+  });
+
+  it("returns correct emoji for .json", () => {
+    expect(getIcon("package.json", false)).toBe("{}");
+  });
+
+  it("returns correct emoji for .html", () => {
+    expect(getIcon("index.html", false)).toBe("🌐");
+  });
+
+  it("returns correct emoji for .css", () => {
+    expect(getIcon("style.css", false)).toBe("🎨");
+  });
+
+  it("returns correct emoji for .sh", () => {
+    expect(getIcon("deploy.sh", false)).toBe("▸");
+  });
+
+  it("returns default file emoji for unknown extensions", () => {
+    expect(getIcon("Makefile", false)).toBe("📄");
+    expect(getIcon("Dockerfile", false)).toBe("📄");
+    expect(getIcon("Rakefile", false)).toBe("📄");
+  });
+
+  it("extension matching is case-insensitive", () => {
+    expect(getIcon("readme.MD", false)).toBe("📄");
+    expect(getIcon("script.PY", false)).toBe("🐍");
+  });
+});
+
+// ─── buildTree ───────────────────────────────────────────────────────────────
+
+describe("buildTree", () => {
+  it("returns empty array for empty input", () => {
+    expect(buildTree([])).toEqual([]);
+  });
+
+  it("adds a single file at root", () => {
+    const files: FileEntry[] = [{ path: "config.yaml", size: 128, dir: false }];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(1);
+    expect(tree[0]).toMatchObject({
+      name: "config.yaml",
+      path: "config.yaml",
+      isDir: false,
+      children: [],
+      size: 128,
+    });
+  });
+
+  it("adds a single directory at root", () => {
+    const files: FileEntry[] = [{ path: "skills", size: 0, dir: true }];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(1);
+    expect(tree[0]).toMatchObject({
+      name: "skills",
+      path: "skills",
+      isDir: true,
+      children: [],
+      size: 0,
+    });
+  });
+
+  it("sorts dirs before files at the same level", () => {
+    const files: FileEntry[] = [
+      { path: "b.txt", size: 10, dir: false },
+      { path: "a.txt", size: 10, dir: false },
+      { path: "z-dir", size: 0, dir: true },
+      { path: "a-dir", size: 0, dir: true },
+    ];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(4);
+    // Dirs first: z-dir, a-dir alphabetically → a before z
+    expect(tree[0].name).toBe("a-dir");
+    expect(tree[1].name).toBe("z-dir");
+    // Then files alphabetically
+    expect(tree[2].name).toBe("a.txt");
+    expect(tree[3].name).toBe("b.txt");
+  });
+
+  it("alphabetically sorts files within the same level", () => {
+    const files: FileEntry[] = [
+      { path: "z.yaml", size: 10, dir: false },
+      { path: "a.yaml", size: 10, dir: false },
+      { path: "m.yaml", size: 10, dir: false },
+    ];
+    const tree = buildTree(files);
+    expect(tree.map((n) => n.name)).toEqual(["a.yaml", "m.yaml", "z.yaml"]);
+  });
+
+  it("nests a file under its parent directory", () => {
+    const files: FileEntry[] = [
+      { path: "skills", size: 0, dir: true },
+      { path: "skills/readme.md", size: 64, dir: false },
+    ];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(1);
+    expect(tree[0].name).toBe("skills");
+    expect(tree[0].children).toHaveLength(1);
+    expect(tree[0].children[0]).toMatchObject({
+      name: "readme.md",
+      path: "skills/readme.md",
+      isDir: false,
+      size: 64,
+    });
+  });
+
+  it("creates intermediate directories automatically", () => {
+    const files: FileEntry[] = [
+      { path: "a/b/c/deep.txt", size: 32, dir: false },
+    ];
+    const tree = buildTree(files);
+    // Root has one child: "a"
+    expect(tree).toHaveLength(1);
+    expect(tree[0].name).toBe("a");
+    expect(tree[0].isDir).toBe(true);
+    // "a" has one child: "b"
+    expect(tree[0].children).toHaveLength(1);
+    expect(tree[0].children[0].name).toBe("b");
+    // "b" has one child: "c"
+    expect(tree[0].children[0].children).toHaveLength(1);
+    expect(tree[0].children[0].children[0].name).toBe("c");
+    // "c" has the file
+    expect(tree[0].children[0].children[0].children[0].name).toBe("deep.txt");
+    expect(tree[0].children[0].children[0].children[0].size).toBe(32);
+  });
+
+  it("adds multiple files to the same directory", () => {
+    const files: FileEntry[] = [
+      { path: "configs", size: 0, dir: true },
+      { path: "configs/a.yaml", size: 10, dir: false },
+      { path: "configs/b.yaml", size: 20, dir: false },
+    ];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(1);
+    expect(tree[0].children.map((n) => n.name).sort()).toEqual(["a.yaml", "b.yaml"]);
+  });
+
+  it("does not duplicate a directory already created as intermediate", () => {
+    const files: FileEntry[] = [
+      { path: "a/b.txt", size: 5, dir: false },
+      { path: "a", size: 0, dir: true },
+    ];
+    const tree = buildTree(files);
+    // "a" should appear only once
+    expect(tree).toHaveLength(1);
+    expect(tree[0].name).toBe("a");
+    // The dir "a" should still contain "b.txt"
+    expect(tree[0].children).toHaveLength(1);
+    expect(tree[0].children[0].name).toBe("b.txt");
+  });
+
+  it("intermediate dirs have size 0", () => {
+    const files: FileEntry[] = [
+      { path: "a/b/c/file.txt", size: 1, dir: false },
+    ];
+    const tree = buildTree(files);
+    expect(tree[0].size).toBe(0);
+    expect(tree[0].children[0].size).toBe(0);
+  });
+
+  it("handles deeply nested mixed dirs and files", () => {
+    const files: FileEntry[] = [
+      { path: "a", size: 0, dir: true },
+      { path: "a/b", size: 0, dir: true },
+      { path: "a/b/c", size: 0, dir: true },
+      { path: "a/b/c/d.txt", size: 1, dir: false },
+      { path: "a/b/e.txt", size: 2, dir: false },
+      { path: "a/f.txt", size: 3, dir: false },
+    ];
+    const tree = buildTree(files);
+    expect(tree).toHaveLength(1); // root: "a"
+    expect(tree[0].children.map((n) => n.name).sort()).toEqual(["b", "f.txt"]);
+    expect(tree[0].children.find((n) => n.name === "b")!.children.map((n) => n.name).sort())
+      .toEqual(["c", "e.txt"]);
+  });
+});

canvas/src/components/tabs/ScheduleTab.tsx

@@ -194,7 +194,7 @@ export function ScheduleTab({ workspaceId }: Props) {
         </span>
         <button
           onClick={() => { resetForm(); setShowForm(true); }}
-          className="text-[11px] px-2 py-0.5 bg-accent-strong/20 text-accent rounded hover:bg-accent-strong/30 transition-colors"
+          className="text-[11px] px-2 py-0.5 bg-accent-strong/20 text-accent rounded hover:bg-accent-strong/30 transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
         >
           + Add Schedule
         </button>
@@ -332,7 +332,14 @@ export function ScheduleTab({ workspaceId }: Props) {
                   <div className="flex items-center gap-1.5">
                     <button
                       onClick={() => handleToggle(sched)}
-                      className={`w-2 h-2 rounded-full flex-shrink-0 ${
+                      aria-label={
+                        sched.last_status === "error"
+                          ? "Last run failed — click to disable"
+                          : sched.last_status === "ok"
+                          ? "Last run OK — click to disable"
+                          : "Never run — click to enable"
+                      }
+                      className={`w-2 h-2 rounded-full flex-shrink-0 focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900 ${
                         sched.last_status === "error"
                           ? "bg-red-400"
                           : sched.last_status === "ok"
@@ -360,7 +367,7 @@ export function ScheduleTab({ workspaceId }: Props) {
                     <span>Runs: {sched.run_count}</span>
                   </div>
                   {sched.last_error && (
-                    <div className="text-[8px] text-bad/70 mt-0.5 truncate">
+                    <div className="text-[8px] text-bad mt-0.5 truncate">
                       Error: {sched.last_error}
                     </div>
                   )}
@@ -369,7 +376,7 @@ export function ScheduleTab({ workspaceId }: Props) {
                   <button
                     onClick={() => handleRunNow(sched)}
                     aria-label={`Run schedule ${sched.name} now`}
-                    className="text-[11px] px-1.5 py-0.5 text-accent hover:bg-accent-strong/20 rounded transition-colors"
+                    className="text-[11px] px-1.5 py-0.5 text-accent hover:bg-accent-strong/20 rounded transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
                     title="Run now"
                   >
                     ▶
@@ -377,7 +384,7 @@ export function ScheduleTab({ workspaceId }: Props) {
                   <button
                     onClick={() => handleEdit(sched)}
                     aria-label={`Edit schedule ${sched.name}`}
-                    className="text-[11px] px-1.5 py-0.5 text-ink-mid hover:bg-surface-card rounded transition-colors"
+                    className="text-[11px] px-1.5 py-0.5 text-ink-mid hover:bg-surface-card rounded transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-accent focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
                     title="Edit"
                   >
                     ✎
@@ -385,7 +392,7 @@ export function ScheduleTab({ workspaceId }: Props) {
                   <button
                     onClick={() => setPendingDelete({ id: sched.id, name: sched.name })}
                     aria-label={`Delete schedule ${sched.name}`}
-                    className="text-[11px] px-1.5 py-0.5 text-bad hover:bg-red-600/20 rounded transition-colors"
+                    className="text-[11px] px-1.5 py-0.5 text-bad hover:bg-red-600/20 rounded transition-colors focus-visible:outline-none focus-visible:ring-2 focus-visible:ring-red-400 focus-visible:ring-offset-1 focus-visible:ring-offset-zinc-900"
                     title="Delete"
                   >
                     ✕

canvas/src/components/tabs/SkillsTab.tsx

@@ -492,7 +492,7 @@ export function SkillsTab({ workspaceId, data }: Props) {
                 <div className="text-[10px] text-bad font-semibold mb-0.5">
                   Couldn't load the plugin registry
                 </div>
-                <div className="text-[10px] text-bad/80">{registryError}</div>
+                <div className="text-[10px] text-bad">{registryError}</div>
                 <div className="mt-1 text-[10px] text-ink-mid">
                   Check the platform server is reachable at /plugins. The Retry button is in the header above.
                 </div>

canvas/src/components/tabs/chat/hooks/index.ts

@@ -0,0 +1,3 @@
+export { useChatHistory } from "./useChatHistory";
+export { useChatSend } from "./useChatSend";
+export { useChatSocket } from "./useChatSocket";

canvas/src/components/tabs/chat/hooks/resolveWorkspaceName.ts

@@ -0,0 +1,11 @@
+"use client";
+
+import { useCanvasStore, type WorkspaceNodeData } from "@/store/canvas";
+
+/** Resolve a workspace ID to its human-readable name.
+ *  Falls back to the first 8 chars of the ID. */
+export function resolveWorkspaceName(id: string): string {
+  const nodes = useCanvasStore.getState().nodes;
+  const node = nodes.find((n) => n.id === id);
+  return (node?.data as WorkspaceNodeData)?.name || id.slice(0, 8);
+}

canvas/src/components/tabs/chat/hooks/useChatHistory.ts

@@ -0,0 +1,134 @@
+"use client";
+
+import { useCallback, useEffect, useRef, useState } from "react";
+import { api } from "@/lib/api";
+import { type ChatMessage, appendMessageDeduped as appendMessageDedupedFn } from "../types";
+
+const INITIAL_HISTORY_LIMIT = 10;
+const OLDER_HISTORY_BATCH = 20;
+
+async function loadMessagesFromDB(
+  workspaceId: string,
+  limit: number,
+  beforeTs?: string,
+): Promise<{ messages: ChatMessage[]; error: string | null; reachedEnd: boolean }> {
+  try {
+    const params = new URLSearchParams({ limit: String(limit) });
+    if (beforeTs) params.set("before_ts", beforeTs);
+    const resp = await api.get<{ messages: ChatMessage[]; reached_end: boolean }>(
+      `/workspaces/${workspaceId}/chat-history?${params.toString()}`,
+    );
+    return {
+      messages: resp.messages ?? [],
+      error: null,
+      reachedEnd: resp.reached_end,
+    };
+  } catch (err) {
+    return {
+      messages: [],
+      error: err instanceof Error ? err.message : "Failed to load chat history",
+      reachedEnd: true,
+    };
+  }
+}
+
+export interface ScrollAnchor {
+  savedDistanceFromBottom: number;
+  expectFirstIdNotEqual: string | null;
+}
+
+export function useChatHistory(
+  workspaceId: string,
+  containerRef?: React.RefObject<HTMLDivElement | null>,
+) {
+  const [messages, setMessages] = useState<ChatMessage[]>([]);
+  const [loading, setLoading] = useState(true);
+  const [loadError, setLoadError] = useState<string | null>(null);
+  const [loadingOlder, setLoadingOlder] = useState(false);
+  const [hasMore, setHasMore] = useState(true);
+
+  const fetchTokenRef = useRef(0);
+  const oldestMessageRef = useRef<ChatMessage | null>(null);
+  const hasMoreRef = useRef(true);
+  const inflightRef = useRef(false);
+  const scrollAnchorRef = useRef<ScrollAnchor | null>(null);
+
+  useEffect(() => {
+    oldestMessageRef.current = messages[0] ?? null;
+  }, [messages]);
+
+  useEffect(() => {
+    hasMoreRef.current = hasMore;
+  }, [hasMore]);
+
+  const loadInitial = useCallback(() => {
+    setLoading(true);
+    setLoadError(null);
+    setHasMore(true);
+    fetchTokenRef.current += 1;
+    const myToken = fetchTokenRef.current;
+    return loadMessagesFromDB(workspaceId, INITIAL_HISTORY_LIMIT).then(
+      ({ messages: msgs, error: fetchErr, reachedEnd }) => {
+        if (fetchTokenRef.current !== myToken) return;
+        setMessages(msgs);
+        setLoadError(fetchErr);
+        setHasMore(!reachedEnd);
+        setLoading(false);
+      },
+    );
+  }, [workspaceId]);
+
+  useEffect(() => {
+    loadInitial();
+  }, [loadInitial]);
+
+  const loadOlder = useCallback(async () => {
+    if (inflightRef.current || !hasMoreRef.current) return;
+    const oldest = oldestMessageRef.current;
+    if (!oldest) return;
+    const container = containerRef?.current;
+    if (!container) return;
+    inflightRef.current = true;
+    scrollAnchorRef.current = {
+      savedDistanceFromBottom: container.scrollHeight - container.scrollTop,
+      expectFirstIdNotEqual: oldest.id,
+    };
+    fetchTokenRef.current += 1;
+    const myToken = fetchTokenRef.current;
+    setLoadingOlder(true);
+    try {
+      const { messages: older, reachedEnd } = await loadMessagesFromDB(
+        workspaceId,
+        OLDER_HISTORY_BATCH,
+        oldest.timestamp,
+      );
+      if (fetchTokenRef.current !== myToken) {
+        scrollAnchorRef.current = null;
+        return;
+      }
+      if (older.length > 0) {
+        setMessages((prev) => [...older, ...prev]);
+      } else {
+        scrollAnchorRef.current = null;
+      }
+      setHasMore(!reachedEnd);
+    } finally {
+      setLoadingOlder(false);
+      inflightRef.current = false;
+    }
+  }, [workspaceId, containerRef]);
+
+  return {
+    messages,
+    loading,
+    loadError,
+    loadingOlder,
+    hasMore,
+    loadInitial,
+    loadOlder,
+    appendMessageDeduped: (msg: ChatMessage) =>
+      setMessages((prev) => appendMessageDedupedFn(prev, msg)),
+    setMessages,
+    scrollAnchorRef,
+  };
+}

canvas/src/components/tabs/chat/hooks/useChatSend.ts

@@ -0,0 +1,182 @@
+"use client";
+
+import { useCallback, useRef, useState } from "react";
+import { api } from "@/lib/api";
+import { uploadChatFiles } from "../uploads";
+import { createMessage, type ChatMessage, type ChatAttachment } from "../types";
+import { extractFilesFromTask } from "../message-parser";
+
+interface A2APart {
+  kind: string;
+  text?: string;
+  file?: {
+    name?: string;
+    mimeType?: string;
+    uri?: string;
+    size?: number;
+  };
+}
+
+interface A2AResponse {
+  result?: {
+    parts?: A2APart[];
+    artifacts?: Array<{ parts: A2APart[] }>;
+  };
+}
+
+export function extractReplyText(resp: A2AResponse): string {
+  const collect = (parts: A2APart[] | undefined): string => {
+    if (!parts) return "";
+    return parts
+      .filter((p) => p.kind === "text")
+      .map((p) => p.text ?? "")
+      .filter(Boolean)
+      .join("\n");
+  };
+  const result = resp?.result;
+  const collected: string[] = [];
+  const fromParts = collect(result?.parts);
+  if (fromParts) collected.push(fromParts);
+  if (result?.artifacts) {
+    for (const a of result.artifacts) {
+      const t = collect(a.parts);
+      if (t) collected.push(t);
+    }
+  }
+  return collected.join("\n");
+}
+
+export interface UseChatSendOptions {
+  getHistoryMessages: () => ChatMessage[];
+  onUserMessage?: (msg: ChatMessage) => void;
+  onAgentMessage?: (msg: ChatMessage) => void;
+}
+
+export function useChatSend(workspaceId: string, options: UseChatSendOptions) {
+  const [sending, setSending] = useState(false);
+  const [uploading, setUploading] = useState(false);
+  const [error, setError] = useState<string | null>(null);
+  const sendInFlightRef = useRef(false);
+  const sendingFromAPIRef = useRef(false);
+  const sendTokenRef = useRef(0);
+  const optionsRef = useRef(options);
+  optionsRef.current = options;
+
+  const releaseSendGuards = useCallback(() => {
+    setSending(false);
+    sendingFromAPIRef.current = false;
+    sendInFlightRef.current = false;
+  }, []);
+
+  const clearError = useCallback(() => setError(null), []);
+
+  const sendMessage = useCallback(
+    async (text: string, files: File[] = []) => {
+      const trimmed = text.trim();
+      if ((!trimmed && files.length === 0) || sending || uploading) return;
+      if (sendInFlightRef.current) return;
+      sendInFlightRef.current = true;
+
+      let uploaded: ChatAttachment[] = [];
+      if (files.length > 0) {
+        setUploading(true);
+        try {
+          uploaded = await uploadChatFiles(workspaceId, files);
+        } catch (e) {
+          setUploading(false);
+          sendInFlightRef.current = false;
+          setError(
+            e instanceof Error ? `Upload failed: ${e.message}` : "Upload failed",
+          );
+          return;
+        }
+        setUploading(false);
+      }
+
+      const userMsg = createMessage("user", trimmed, uploaded);
+      optionsRef.current.onUserMessage?.(userMsg);
+
+      setSending(true);
+      sendingFromAPIRef.current = true;
+      setError(null);
+      const myToken = ++sendTokenRef.current;
+
+      const history = optionsRef.current
+        .getHistoryMessages()
+        .filter((m) => m.role === "user" || m.role === "agent")
+        .slice(-20)
+        .map((m) => ({
+          role: m.role === "user" ? "user" : "agent",
+          parts: [{ kind: "text", text: m.content }],
+        }));
+
+      const parts: A2APart[] = [];
+      if (trimmed) parts.push({ kind: "text", text: trimmed });
+      for (const att of uploaded) {
+        parts.push({
+          kind: "file",
+          file: {
+            name: att.name,
+            mimeType: att.mimeType,
+            uri: att.uri,
+            size: att.size,
+          },
+        });
+      }
+
+      api
+        .post<A2AResponse>(
+          `/workspaces/${workspaceId}/a2a`,
+          {
+            method: "message/send",
+            params: {
+              message: {
+                role: "user",
+                messageId: crypto.randomUUID(),
+                parts,
+              },
+              metadata: { history },
+            },
+          },
+          { timeoutMs: 120_000 },
+        )
+        .then((resp) => {
+          if (sendTokenRef.current !== myToken) return;
+          if (!sendingFromAPIRef.current) {
+            sendInFlightRef.current = false;
+            return;
+          }
+          const replyText = extractReplyText(resp);
+          const replyFiles = extractFilesFromTask(
+            (resp?.result ?? {}) as Record<string, unknown>,
+          );
+          if (replyText || replyFiles.length > 0) {
+            optionsRef.current.onAgentMessage?.(
+              createMessage("agent", replyText, replyFiles),
+            );
+          }
+          releaseSendGuards();
+        })
+        .catch(() => {
+          if (sendTokenRef.current !== myToken) return;
+          if (!sendingFromAPIRef.current) {
+            sendInFlightRef.current = false;
+            return;
+          }
+          releaseSendGuards();
+          setError("Failed to send message — agent may be unreachable");
+        });
+    },
+    [workspaceId, sending, uploading],
+  );
+
+  return {
+    sending,
+    uploading,
+    sendMessage,
+    error,
+    clearError,
+    releaseSendGuards,
+    sendingFromAPIRef,
+  };
+}

canvas/src/components/tabs/chat/hooks/useChatSocket.ts

@@ -0,0 +1,100 @@
+"use client";
+
+import { useCallback, useEffect, useRef } from "react";
+import { useCanvasStore, type WorkspaceNodeData } from "@/store/canvas";
+import { useSocketEvent } from "@/hooks/useSocketEvent";
+import { createMessage, type ChatMessage } from "../types";
+
+export interface UseChatSocketCallbacks {
+  onAgentMessage?: (msg: ChatMessage) => void;
+  onActivityLog?: (entry: string) => void;
+  onSendComplete?: () => void;
+  onSendError?: (error: string) => void;
+}
+
+export function useChatSocket(
+  workspaceId: string,
+  callbacks: UseChatSocketCallbacks,
+): void {
+  const callbacksRef = useRef(callbacks);
+  callbacksRef.current = callbacks;
+
+  // Agent push messages from global store
+  const pendingAgentMsgs = useCanvasStore((s) => s.agentMessages[workspaceId]);
+  useEffect(() => {
+    if (!pendingAgentMsgs || pendingAgentMsgs.length === 0) return;
+    const consume = useCanvasStore.getState().consumeAgentMessages;
+    const msgs = consume(workspaceId);
+    for (const m of msgs) {
+      callbacksRef.current.onAgentMessage?.(
+        createMessage("agent", m.content, m.attachments),
+      );
+    }
+    if (msgs.length > 0) {
+      callbacksRef.current.onSendComplete?.();
+    }
+  }, [pendingAgentMsgs, workspaceId]);
+
+  const resolveWorkspaceName = useCallback((id: string) => {
+    const nodes = useCanvasStore.getState().nodes;
+    const node = nodes.find((n) => n.id === id);
+    return (node?.data as WorkspaceNodeData)?.name || id.slice(0, 8);
+  }, []);
+
+  useSocketEvent((msg) => {
+    try {
+      if (msg.event === "ACTIVITY_LOGGED") {
+        if (msg.workspace_id !== workspaceId) return;
+
+        const p = msg.payload || {};
+        const type = p.activity_type as string;
+        const method = (p.method as string) || "";
+        const status = (p.status as string) || "";
+        const targetId = (p.target_id as string) || "";
+        const durationMs = p.duration_ms as number | undefined;
+        const summary = (p.summary as string) || "";
+
+        let line = "";
+        if (type === "a2a_receive" && method === "message/send") {
+          const targetName = resolveWorkspaceName(targetId || msg.workspace_id);
+          if (status === "ok" && durationMs) {
+            const sec = Math.round(durationMs / 1000);
+            line = `← ${targetName} responded (${sec}s)`;
+            const own = (targetId || msg.workspace_id) === workspaceId;
+            if (own) callbacksRef.current.onSendComplete?.();
+          } else if (status === "error") {
+            line = `⚠ ${targetName} error`;
+            const own = (targetId || msg.workspace_id) === workspaceId;
+            if (own) {
+              callbacksRef.current.onSendComplete?.();
+              callbacksRef.current.onSendError?.(
+                "Agent error (Exception) — see workspace logs for details.",
+              );
+            }
+          }
+        } else if (type === "a2a_send") {
+          const targetName = resolveWorkspaceName(targetId);
+          line = `→ Delegating to ${targetName}...`;
+        } else if (type === "task_update") {
+          if (summary) line = `⟳ ${summary}`;
+        } else if (type === "agent_log") {
+          if (summary) line = summary;
+        }
+
+        if (line) {
+          callbacksRef.current.onActivityLog?.(line);
+        }
+      } else if (
+        msg.event === "TASK_UPDATED" &&
+        msg.workspace_id === workspaceId
+      ) {
+        const task = (msg.payload?.current_task as string) || "";
+        if (task) {
+          callbacksRef.current.onActivityLog?.(`⟳ ${task}`);
+        }
+      }
+    } catch {
+      /* ignore */
+    }
+  });
+}

canvas/src/components/tabs/chat/index.ts

@@ -1,2 +1,5 @@
 export { type ChatMessage, createMessage, appendMessageDeduped } from "./types";
 export { extractAgentText, extractTextsFromParts, extractResponseText } from "./message-parser";
+export { useChatHistory } from "./hooks/useChatHistory";
+export { useChatSend } from "./hooks/useChatSend";
+export { useChatSocket } from "./hooks/useChatSocket";

canvas/src/hooks/useTemplateDeploy.tsx

@@ -8,6 +8,7 @@ import {
   type PreflightResult,
   type Template,
 } from "@/lib/deploy-preflight";
+import { isSaaSTenant } from "@/lib/tenant";
 import { MissingKeysModal } from "@/components/MissingKeysModal";
 
 /**
@@ -105,7 +106,7 @@ export function useTemplateDeploy(
         const ws = await api.post<{ id: string }>("/workspaces", {
           name: template.name,
           template: template.id,
-          tier: template.tier,
+          tier: isSaaSTenant() ? 4 : template.tier,
           canvas: coords,
           ...(model ? { model } : {}),
         });

canvas/src/lib/__tests__/hydrate.test.ts

@@ -0,0 +1,189 @@
+// @vitest-environment jsdom
+/**
+ * Tests for hydrate.ts — canvas store hydration with exponential backoff.
+ *
+ * Covers:
+ *   - Successful hydration on first attempt (no retries)
+ *   - Retry with exponential backoff on failure
+ *   - onRetrying callback called at correct intervals
+ *   - Error propagation after MAX_RETRIES exhausted
+ *   - Viewport persisted on success
+ *   - Viewport failure is non-fatal
+ */
+import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
+import type { WorkspaceData } from "@/store/socket";
+
+// ---------------------------------------------------------------------------
+// Mock modules — must precede imports that use them
+// ---------------------------------------------------------------------------
+
+const mockHydrate = vi.fn();
+const mockSetViewport = vi.fn();
+
+vi.mock("@/lib/api", () => ({
+  api: {
+    get: vi.fn(),
+  },
+  PLATFORM_URL: "https://platform.test",
+}));
+
+vi.mock("@/store/canvas", () => ({
+  useCanvasStore: Object.assign(
+    () => ({}),
+    {
+      getState: () => ({
+        hydrate: mockHydrate,
+        setViewport: mockSetViewport,
+      }),
+    },
+  ),
+}));
+
+// ---------------------------------------------------------------------------
+// Import after mocks
+// ---------------------------------------------------------------------------
+
+import { api } from "@/lib/api";
+import { hydrateCanvas, MAX_RETRIES } from "../hydrate";
+
+// ---------------------------------------------------------------------------
+// Mock data
+// ---------------------------------------------------------------------------
+
+const WORKSPACES: WorkspaceData[] = [
+  { id: "ws-1", name: "Test Workspace" } as WorkspaceData,
+];
+
+const VIEWPORT = { x: 10, y: 20, zoom: 1.5 };
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+const mockApiGet = vi.mocked(api.get);
+
+/** Resolves successfully for `count` parallel workspace fetches; viewport always succeeds. */
+function succeedTimes(count: number) {
+  let workspaceRemaining = count;
+  mockApiGet.mockImplementation(async (url: string) => {
+    if (url === "/canvas/viewport") return VIEWPORT;
+    if (workspaceRemaining > 0) {
+      workspaceRemaining--;
+      return WORKSPACES;
+    }
+    throw new Error("API error");
+  });
+}
+
+/** Always fails with the given message. */
+function alwaysFail(msg = "Network error") {
+  mockApiGet.mockRejectedValue(new Error(msg));
+}
+
+// ---------------------------------------------------------------------------
+// Tests
+// ---------------------------------------------------------------------------
+
+describe("hydrateCanvas", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+    mockApiGet.mockReset();
+    mockHydrate.mockReset();
+    mockSetViewport.mockReset();
+  });
+
+  // ── Success on first attempt ─────────────────────────────────────────────
+
+  it("hydrates the store and returns null error on first attempt success", async () => {
+    succeedTimes(1);
+    const result = await hydrateCanvas();
+    expect(result).toEqual({ error: null });
+    expect(mockHydrate).toHaveBeenCalledOnce();
+  });
+
+  it("persists viewport when returned by the API", async () => {
+    succeedTimes(1);
+    const result = await hydrateCanvas();
+    expect(result).toEqual({ error: null });
+    expect(mockSetViewport).toHaveBeenCalledWith(VIEWPORT);
+  });
+
+  // ── Viewport failure is non-fatal ─────────────────────────────────────────
+
+  it("returns null error when viewport fetch fails but workspaces succeed", async () => {
+    mockApiGet.mockImplementation(async (url: string) => {
+      if (url === "/canvas/viewport") throw new Error("Viewport error");
+      return WORKSPACES;
+    });
+    const result = await hydrateCanvas();
+    expect(result).toEqual({ error: null });
+    expect(mockHydrate).toHaveBeenCalledOnce();
+    expect(mockSetViewport).not.toHaveBeenCalled();
+  });
+
+  // ── Retry logic ──────────────────────────────────────────────────────────
+
+  it("retries MAX_RETRIES times before returning an error", async () => {
+    alwaysFail();
+    const onRetrying = vi.fn();
+    const result = await Promise.race([
+      hydrateCanvas(onRetrying),
+      new Promise<"timeout">((resolve) => setTimeout(() => resolve("timeout"), 5000)),
+    ]);
+    if (result === "timeout") throw new Error("Test timed out — retries not awaited correctly");
+    expect(result.error).not.toBeNull();
+    expect(onRetrying).toHaveBeenCalledTimes(MAX_RETRIES - 1);
+  }, 10000);
+
+  it("onRetrying is called with attempt number before each retry", async () => {
+    alwaysFail();
+    const onRetrying = vi.fn();
+    await Promise.race([
+      hydrateCanvas(onRetrying),
+      new Promise<"timeout">((resolve) => setTimeout(() => resolve("timeout"), 5000)),
+    ]);
+    expect(onRetrying).toHaveBeenNthCalledWith(1, 1);
+    expect(onRetrying).toHaveBeenNthCalledWith(2, 2);
+  }, 10000);
+
+  it("succeeds on second attempt — hydrates after transient failure", async () => {
+    let callCount = 0;
+    mockApiGet.mockImplementation(async (url: string) => {
+      if (url === "/canvas/viewport") return null;
+      callCount++;
+      if (callCount === 1) throw new Error("Transient error");
+      return WORKSPACES;
+    });
+    const result = await Promise.race([
+      hydrateCanvas(),
+      new Promise<"timeout">((resolve) => setTimeout(() => resolve("timeout"), 5000)),
+    ]);
+    if (result === "timeout") throw new Error("Test timed out");
+    expect(result).toEqual({ error: null });
+    expect(mockHydrate).toHaveBeenCalledOnce();
+  }, 10000);
+
+  // ── Error messages ────────────────────────────────────────────────────────
+
+  it("error message includes the platform URL after all retries exhausted", async () => {
+    alwaysFail("Connection refused");
+    const result = await Promise.race([
+      hydrateCanvas(),
+      new Promise<"timeout">((resolve) => setTimeout(() => resolve("timeout"), 5000)),
+    ]);
+    if (result === "timeout") throw new Error("Test timed out");
+    expect(result.error).toContain("platform.test");
+    expect(result.error).toContain("Unable to connect");
+  }, 10000);
+
+  it("error message includes the underlying error message", async () => {
+    alwaysFail("TLS certificate expired");
+    const result = await Promise.race([
+      hydrateCanvas(),
+      new Promise<"timeout">((resolve) => setTimeout(() => resolve("timeout"), 5000)),
+    ]);
+    if (result === "timeout") throw new Error("Test timed out");
+    expect(result.error).not.toBeNull();
+    expect(typeof result.error).toBe("string");
+  }, 10000);
+});

canvas/src/lib/__tests__/palette-context.test.tsx

@@ -0,0 +1,205 @@
+// @vitest-environment jsdom
+"use client";
+/**
+ * Tests for palette-context.tsx — MobileAccentProvider context + usePalette hook.
+ *
+ * Test coverage (9 cases):
+ * 1. MobileAccentProvider renders children
+ * 2. usePalette(false) without provider → MOL_LIGHT
+ * 3. usePalette(true) without provider → MOL_DARK
+ * 4. accent=null returns base palette unchanged
+ * 5. accent=base.accent returns base palette unchanged (identity guard)
+ * 6. accent="#custom" overrides both accent and online
+ * 7. MOL_LIGHT singleton never mutated
+ * 8. MOL_DARK singleton never mutated
+ *
+ * Plus pure-function coverage for normalizeStatus + tierCode.
+ */
+import { describe, expect, it, vi, beforeEach, afterEach } from "vitest";
+import React from "react";
+import { render, screen, cleanup } from "@testing-library/react";
+import {
+  MOL_LIGHT,
+  MOL_DARK,
+  getPalette,
+  normalizeStatus,
+  tierCode,
+  MobileAccentProvider,
+  usePalette,
+} from "../palette-context";
+
+// ─── usePalette test helper ───────────────────────────────────────────────────
+// usePalette reads document.documentElement.dataset.theme internally.
+// We set this before rendering so the hook sees the right value.
+
+function setDataTheme(theme: "light" | "dark") {
+  if (typeof document !== "undefined") {
+    document.documentElement.dataset.theme = theme;
+  }
+}
+
+// ─── Pure function tests ──────────────────────────────────────────────────────
+
+describe("normalizeStatus", () => {
+  it("returns emerald-400 for online status", () => {
+    expect(normalizeStatus("online", false)).toBe("bg-emerald-400");
+    expect(normalizeStatus("online", true)).toBe("bg-emerald-400");
+  });
+
+  it("returns emerald-400 for degraded status", () => {
+    expect(normalizeStatus("degraded", false)).toBe("bg-emerald-400");
+    expect(normalizeStatus("degraded", true)).toBe("bg-emerald-400");
+  });
+
+  it("returns red-400 for failed status", () => {
+    expect(normalizeStatus("failed", false)).toBe("bg-red-400");
+    expect(normalizeStatus("failed", true)).toBe("bg-red-400");
+  });
+
+  it("returns amber-400 for paused status", () => {
+    expect(normalizeStatus("paused", false)).toBe("bg-amber-400");
+    expect(normalizeStatus("paused", true)).toBe("bg-amber-400");
+  });
+
+  it("returns amber-400 for not_configured status", () => {
+    expect(normalizeStatus("not_configured", false)).toBe("bg-amber-400");
+  });
+
+  it("returns zinc-400 for unknown status", () => {
+    expect(normalizeStatus("unknown", false)).toBe("bg-zinc-400");
+    expect(normalizeStatus("", false)).toBe("bg-zinc-400");
+  });
+});
+
+describe("tierCode", () => {
+  it("returns T1 for tier 1", () => {
+    expect(tierCode(1)).toBe("T1");
+  });
+
+  it("returns T2 for tier 2", () => {
+    expect(tierCode(2)).toBe("T2");
+  });
+
+  it("returns T4 for tier 4", () => {
+    expect(tierCode(4)).toBe("T4");
+  });
+
+  it("returns generic T{n} for non-standard tiers", () => {
+    expect(tierCode(99)).toBe("T99");
+  });
+});
+
+// ─── getPalette tests ─────────────────────────────────────────────────────────
+
+describe("getPalette — accent override", () => {
+  it("accent=null returns base palette unchanged (light)", () => {
+    const result = getPalette(null, false);
+    expect(result).toEqual({ ...MOL_LIGHT });
+    expect(result).not.toBe(MOL_LIGHT); // returned object is a copy
+  });
+
+  it("accent=null returns base palette unchanged (dark)", () => {
+    const result = getPalette(null, true);
+    expect(result).toEqual({ ...MOL_DARK });
+    expect(result).not.toBe(MOL_DARK);
+  });
+
+  it("accent=base.accent returns base palette unchanged (identity guard, light)", () => {
+    const result = getPalette(MOL_LIGHT.accent, false);
+    expect(result).toEqual({ ...MOL_LIGHT });
+    expect(result).not.toBe(MOL_LIGHT);
+  });
+
+  it("accent=base.accent returns base palette unchanged (identity guard, dark)", () => {
+    const result = getPalette(MOL_DARK.accent, true);
+    expect(result).toEqual({ ...MOL_DARK });
+    expect(result).not.toBe(MOL_DARK);
+  });
+
+  it("accent='#custom' overrides accent and online (light)", () => {
+    const result = getPalette("#ff0000", false);
+    expect(result.accent).toBe("#ff0000");
+    expect(result.online).toBe("bg-emerald-400"); // normalizeStatus("online", false)
+  });
+
+  it("accent='#custom' overrides accent and online (dark)", () => {
+    const result = getPalette("#00ff00", true);
+    expect(result.accent).toBe("#00ff00");
+    expect(result.online).toBe("bg-emerald-400"); // normalizeStatus("online", true)
+  });
+
+  it("MOL_LIGHT singleton is never mutated", () => {
+    getPalette("#mutate", false);
+    // All fields must still match the original freeze definition
+    expect(MOL_LIGHT.accent).toBe("bg-blue-500");
+    expect(MOL_LIGHT.online).toBe("bg-emerald-400");
+    expect(MOL_LIGHT.surface).toBe("bg-zinc-900");
+    expect(MOL_LIGHT.ink).toBe("text-zinc-100");
+    expect(MOL_LIGHT.line).toBe("border-zinc-700");
+    expect(MOL_LIGHT.bg).toBe("bg-zinc-950");
+  });
+
+  it("MOL_DARK singleton is never mutated", () => {
+    getPalette("#mutate", true);
+    expect(MOL_DARK.accent).toBe("bg-sky-400");
+    expect(MOL_DARK.online).toBe("bg-emerald-400");
+    expect(MOL_DARK.surface).toBe("bg-zinc-800");
+    expect(MOL_DARK.ink).toBe("text-zinc-100");
+    expect(MOL_DARK.line).toBe("border-zinc-700");
+    expect(MOL_DARK.bg).toBe("bg-zinc-950");
+  });
+
+  it("getPalette always returns a new object (no shared mutation risk)", () => {
+    const a = getPalette("#a", false);
+    const b = getPalette("#b", false);
+    expect(a).not.toBe(b);
+    expect(a.accent).not.toBe(b.accent);
+  });
+});
+
+// ─── MobileAccentProvider tests ───────────────────────────────────────────────
+
+describe("MobileAccentProvider", () => {
+  beforeEach(() => {
+    setDataTheme("light");
+  });
+
+  afterEach(() => {
+    cleanup();
+    if (typeof document !== "undefined") {
+      document.documentElement.dataset.theme = "";
+    }
+  });
+
+  it("renders children", () => {
+    render(
+      <MobileAccentProvider accent={null}>
+        <span data-testid="child">Hello</span>
+      </MobileAccentProvider>,
+    );
+    expect(screen.getByTestId("child")).toBeTruthy();
+  });
+
+  // usePalette hook reads data-theme from <html> to determine light/dark.
+  // In the test environment, data-theme is empty, which falls through to
+  // the "light" default in usePalette, giving MOL_LIGHT.
+  it("usePalette(false) without provider → MOL_LIGHT", () => {
+    setDataTheme("light");
+    function ShowPalette() {
+      const p = usePalette(false);
+      return <span data-testid="accent-light">{p.accent}</span>;
+    }
+    render(<ShowPalette />);
+    expect(screen.getByTestId("accent-light").textContent).toBe(MOL_LIGHT.accent);
+  });
+
+  it("usePalette(true) without provider → MOL_DARK when data-theme=dark", () => {
+    setDataTheme("dark");
+    function ShowPalette() {
+      const p = usePalette(true);
+      return <span data-testid="accent-dark">{p.accent}</span>;
+    }
+    render(<ShowPalette />);
+    expect(screen.getByTestId("accent-dark").textContent).toBe(MOL_DARK.accent);
+  });
+});

canvas/src/lib/api.ts

@@ -8,14 +8,18 @@ import { getTenantSlug } from "./tenant";
 export const PLATFORM_URL =
   process.env.NEXT_PUBLIC_PLATFORM_URL ?? "http://localhost:8080";
 
-// 15s is long enough for slow CP queries but short enough that a
-// hung backend doesn't leave the UI spinning forever. The abort
-// propagates through AbortController so React components can observe
-// the error and render a retry affordance. Callers that know the
-// endpoint is intentionally slow (org import walks a tree of
-// workspaces with server-side pacing) can pass `timeoutMs` to
-// override.
-const DEFAULT_TIMEOUT_MS = 15_000;
+// 35s is long enough for the slowest server-side path (EIC SSH
+// tunnel for tenant EC2 file operations, bounded server-side by
+// `eicFileOpTimeout = 30 * time.Second` in
+// workspace-server/internal/handlers/template_files_eic.go) so the
+// canvas surfaces the server's real error instead of aborting first
+// with a generic timeout. Shorter values caused "Save & Restart" to
+// time out at the client before the backend returned its 5xx. The
+// abort still propagates through AbortController so React components
+// can render a retry affordance. Callers that know an endpoint is
+// intentionally slow (org import walks a tree of workspaces with
+// server-side pacing) can pass `timeoutMs` to override.
+const DEFAULT_TIMEOUT_MS = 35_000;
 
 export interface RequestOptions {
   timeoutMs?: number;

canvas/src/lib/design-tokens.ts

@@ -21,8 +21,8 @@ export function statusDotClass(status: string): string {
 export const TIER_CONFIG: Record<number, { label: string; color: string; border: string }> = {
   1: { label: "T1", color: "text-ink-mid bg-surface-card border border-line", border: "text-ink-mid border-line" },
   2: { label: "T2", color: "text-white bg-accent border border-accent-strong", border: "text-accent border-accent" },
-  3: { label: "T3", color: "text-white bg-violet-600 border border-violet-700", border: "text-violet-600 border-violet-500" },
-  4: { label: "T4", color: "text-white bg-warm border border-warm", border: "text-warm border-warm" },
+  3: { label: "T3", color: "text-white bg-violet-600 border border-violet-700", border: "text-white border-violet-500" },
+  4: { label: "T4", color: "text-white bg-warm border border-warm", border: "text-white border-warm" },
 };
 
 export const COMM_TYPE_LABELS: Record<string, string> = {

canvas/src/lib/palette-context.tsx

@@ -0,0 +1,167 @@
+"use client";
+
+/**
+ * palette-context.tsx
+ *
+ * Mobile canvas accent palette system.
+ *
+ * - MOL_LIGHT / MOL_DARK  — immutable base singletons
+ * - getPalette(accent, isDark) — returns base palette or accent-overridden copy
+ * - normalizeStatus(status, isDark) — maps workspace status → online dot color
+ * - tierCode(tier) — maps tier number → display label
+ * - MobileAccentProvider — React context that propagates accent override
+ * - usePalette(allowAccentOverride) — hook; returns the effective palette
+ */
+
+import { createContext, useContext } from "react";
+
+// ─── Types ─────────────────────────────────────────────────────────────────────
+
+export interface Palette {
+  /** Accent colour (CSS colour string). */
+  accent: string;
+  /** Online indicator colour (CSS class string, e.g. "bg-emerald-400"). */
+  online: string;
+  /** Surface background colour class. */
+  surface: string;
+  /** Primary text colour class. */
+  ink: string;
+  /** Border/divider colour class. */
+  line: string;
+  /** Background colour class. */
+  bg: string;
+  /** Tier display code, e.g. "T1". */
+  tier: string;
+}
+
+// ─── Singleton base palettes ────────────────────────────────────────────────────
+
+/** Light-mode base palette — must never be mutated. */
+export const MOL_LIGHT: Readonly<Palette> = Object.freeze({
+  accent: "bg-blue-500",
+  online: "bg-emerald-400",
+  surface: "bg-zinc-900",
+  ink: "text-zinc-100",
+  line: "border-zinc-700",
+  bg: "bg-zinc-950",
+  tier: "T1",
+});
+
+/** Dark-mode base palette — must never be mutated. */
+export const MOL_DARK: Readonly<Palette> = Object.freeze({
+  accent: "bg-sky-400",
+  online: "bg-emerald-400",
+  surface: "bg-zinc-800",
+  ink: "text-zinc-100",
+  line: "border-zinc-700",
+  bg: "bg-zinc-950",
+  tier: "T1",
+});
+
+// ─── Pure helpers ─────────────────────────────────────────────────────────────
+
+/**
+ * Maps workspace status string → online dot colour class.
+ * Returns the appropriate green for light/dark mode.
+ */
+export function normalizeStatus(
+  status: string,
+  _isDark: boolean,
+): string {
+  if (status === "online" || status === "degraded") {
+    return "bg-emerald-400";
+  }
+  if (status === "failed") {
+    return "bg-red-400";
+  }
+  if (status === "paused" || status === "not_configured") {
+    return "bg-amber-400";
+  }
+  return "bg-zinc-400";
+}
+
+/**
+ * Maps tier number → display code.
+ */
+export function tierCode(tier: number): string {
+  return `T${tier}`;
+}
+
+/**
+ * Returns the effective palette.
+ *
+ * - `accent = null` → base palette (light or dark) unchanged
+ * - `accent = basePalette.accent` → base palette unchanged (identity guard)
+ * - `accent = "#custom"` → copy with `accent` and `online` overridden
+ *
+ * Always returns a new object; neither MOL_LIGHT nor MOL_DARK is ever mutated.
+ */
+export function getPalette(
+  accent: string | null,
+  isDark: boolean,
+): Palette {
+  const base: Readonly<Palette> = isDark ? MOL_DARK : MOL_LIGHT;
+
+  // null accent → use base unchanged
+  if (accent === null) return { ...base };
+
+  // identity guard — accent same as base accent → no override needed
+  if (accent === base.accent) return { ...base };
+
+  // Custom accent: override accent + online to keep them in sync
+  return { ...base, accent, online: normalizeStatus("online", isDark) };
+}
+
+// ─── Context ──────────────────────────────────────────────────────────────────
+
+type MobileAccentContextValue = {
+  /** Override accent colour (null = no override, use default). */
+  accent: string | null;
+};
+
+const MobileAccentContext = createContext<MobileAccentContextValue>({
+  accent: null,
+});
+
+export { MobileAccentContext };
+
+/**
+ * Renders children inside the accent override context.
+ */
+export function MobileAccentProvider({
+  accent,
+  children,
+}: {
+  accent: string | null;
+  children: React.ReactNode;
+}) {
+  return (
+    <MobileAccentContext.Provider value={{ accent }}>
+      {children}
+    </MobileAccentContext.Provider>
+  );
+}
+
+// ─── Hook ─────────────────────────────────────────────────────────────────────
+
+/**
+ * Returns the effective `Palette` for the current context.
+ *
+ * @param allowAccentOverride  When false, always returns the base palette
+ *                              even when an override is set (useful for
+ *                              non-accent-aware child components).
+ */
+export function usePalette(allowAccentOverride: boolean): Palette {
+  const { accent } = useContext(MobileAccentContext);
+
+  // Resolved from the OS-level theme preference. In a real app this would
+  // be derived from useTheme().resolvedTheme; for this hook we default
+  // to light (the safe default for SSR / component-library use).
+  // We read data-theme from <html> to stay in sync with the theme system.
+  const isDark =
+    typeof document !== "undefined" &&
+    document.documentElement.dataset.theme === "dark";
+
+  const effectiveAccent = allowAccentOverride ? accent : null;
+  return getPalette(effectiveAccent, isDark);
+}

canvas/src/store/canvas-topology.ts

@@ -519,6 +519,10 @@ export function buildNodesAndEdges(
         // #2054 — server-declared per-workspace provisioning timeout.
         // Falls through to the runtime profile when null/absent.
         provisionTimeoutMs: ws.provision_timeout_ms ?? null,
+        // Workspace abilities — defaults preserved for old platform versions
+        // that don't yet include these columns in the GET response.
+        broadcastEnabled: ws.broadcast_enabled ?? false,
+        talkToUserEnabled: ws.talk_to_user_enabled ?? true,
       },
     };
     if (hasParent) {

canvas/src/store/canvas.ts

@@ -99,6 +99,13 @@ export interface WorkspaceNodeData extends Record<string, unknown> {
    *  @/lib/runtimeProfiles. Lets a slow runtime declare its cold-boot
    *  expectation without a canvas release. */
   provisionTimeoutMs?: number | null;
+  /** When true the workspace may POST /broadcast to send org-wide messages.
+   *  Default false. Toggled by user/admin via PATCH /workspaces/:id/abilities. */
+  broadcastEnabled?: boolean;
+  /** When false the workspace cannot deliver canvas chat messages.
+   *  send_message_to_user / POST /notify return 403 and the canvas
+   *  shows a "not enabled" state with a button to re-enable. Default true. */
+  talkToUserEnabled?: boolean;
 }
 
 export type PanelTab = "details" | "skills" | "chat" | "terminal" | "config" | "schedule" | "channels" | "files" | "memory" | "traces" | "events" | "activity" | "audit";

canvas/src/store/socket.ts

@@ -299,6 +299,9 @@ export interface WorkspaceData {
    *  `@/lib/runtimeProfiles` when absent (the default behavior for any
    *  template that hasn't yet declared the field). */
   provision_timeout_ms?: number | null;
+  /** Workspace ability flags (migration 20260514). */
+  broadcast_enabled?: boolean;
+  talk_to_user_enabled?: boolean;
 }
 
 let socket: ReconnectingSocket | null = null;

canvas/src/styles/settings-panel.css

@@ -282,13 +282,17 @@
 }
 
 .secret-row__save-btn {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 6px 12px;
   border-radius: 6px;
   font-size: 13px;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.secret-row__save-btn:hover {
+  background: #1e40af;
 }
 
 .secret-row__save-btn:focus-visible {
@@ -370,13 +374,17 @@
 }
 
 .add-key-form__save-btn {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 8px 16px;
   border-radius: 6px;
   font-size: 13px;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.add-key-form__save-btn:hover {
+  background: #1e40af;
 }
 
 .add-key-form__save-btn:focus-visible {
@@ -510,7 +518,7 @@
 .empty-state__body { font-size: 14px; color: #a1a1aa; margin: 0 0 24px; line-height: 1.5; }
 
 .empty-state__cta {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 10px 20px;
@@ -518,6 +526,10 @@
   font-size: 14px;
   font-weight: 500;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.empty-state__cta:hover {
+  background: #1e40af;
 }
 
 .empty-state__cta:focus-visible { outline: var(--focus-ring); outline-offset: var(--focus-ring-offset); }
@@ -561,12 +573,16 @@
 .secrets-tab__error p { color: var(--status-invalid); margin: 0 0 12px; }
 
 .secrets-tab__refresh-btn {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 8px 16px;
   border-radius: 6px;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.secrets-tab__refresh-btn:hover {
+  background: #1e40af;
 }
 
 .secrets-tab__no-results {
@@ -690,12 +706,16 @@
 }
 
 .guard-dialog__discard-btn {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 8px 16px;
   border-radius: 6px;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.guard-dialog__discard-btn:hover {
+  background: #1e40af;
 }
 
 .guard-dialog__discard-btn:focus-visible {
@@ -747,12 +767,20 @@
 .top-bar__name { font-size: 14px; font-weight: 500; color: #d4d4d8; }
 
 .top-bar__btn {
-  background: #2563eb;
+  background: #1d4ed8;
   color: #ffffff;
   border: none;
   padding: 6px 12px;
   border-radius: 6px;
   font-size: 13px;
   cursor: pointer;
+  transition: background-color 0.15s;
+}
+.top-bar__btn:hover {
+  background: #1e40af;
+}
+.top-bar__btn:focus-visible {
+  outline: none;
+  box-shadow: 0 0 0 2px #18181b, 0 0 0 4px #3b82f6;
 }
 

scripts/promote-tenant-image.sh

@@ -179,6 +179,7 @@ cp_redeploy_tenant() {
   #   1  — any other failure
   # stdout = response body. stderr = "HTTP_STATUS=NNN" line.
   local slug="$1" tag="$2"
+  validate_slug "$slug"
   _mock_call cp_redeploy_tenant "$slug" "$tag"; local _mrc=$?
   [[ $_mrc -ne 99 ]] && return $_mrc
   local tok="${!CP_TOKEN_ENV:-}"
@@ -204,6 +205,7 @@ cp_redeploy_tenant() {
 tenant_buildinfo() {
   # args: <slug>; prints JSON
   local slug="$1"
+  validate_slug "$slug"
   _mock_call tenant_buildinfo "$slug"; local _mrc=$?
   [[ $_mrc -ne 99 ]] && return $_mrc
   curl -sf --max-time 10 "https://${slug}.moleculesai.app/buildinfo"
@@ -212,6 +214,7 @@ tenant_buildinfo() {
 tenant_health() {
   # args: <slug>; prints raw response, returns 0 if "ok"
   local slug="$1"
+  validate_slug "$slug"
   _mock_call tenant_health "$slug"; local _mrc=$?
   [[ $_mrc -ne 99 ]] && return $_mrc
   curl -sf --max-time 10 "https://${slug}.moleculesai.app/health"
@@ -256,6 +259,7 @@ print(json.dumps({'commands': [ecr_login]}))
 resolve_tenant_instance_id() {
   # args: <slug>; prints i-xxx
   local slug="$1"
+  validate_slug "$slug"
   _mock_call resolve_tenant_instance_id "$slug"; local _mrc=$?
   [[ $_mrc -ne 99 ]] && return $_mrc
   local tok="${!CP_TOKEN_ENV:-}"
@@ -271,6 +275,19 @@ resolve_tenant_instance_id() {
 log() { printf '[%s] %s\n' "$(date -u +%H:%M:%SZ)" "$*"; }
 err() { printf '[%s] ERROR: %s\n' "$(date -u +%H:%M:%SZ)" "$*" >&2; }
 
+# validate_slug — exit 64 if slug contains characters outside the safe set.
+# Prevents SSRF via query-separator injection (?foo) and subdomain takeover
+# (@evil) when slug is interpolated into URL paths or subdomains.
+# OFFSEC-006 fix.
+validate_slug() {
+  local slug="$1"
+  if ! [[ "$slug" =~ ^[a-z0-9]([a-z0-9-]{0,61}[a-z0-9])?$ ]]; then
+    printf '[%s] ERROR: invalid slug: %s\n' \
+      "$(date -u +%H:%M:%SZ)" "$slug" >&2
+    exit 64
+  fi
+}
+
 preflight() {
   log "preflight: source=$SOURCE_TAG dest=$DEST_TAG repo=$REPO region=$REGION"
   local src_manifest
@@ -339,6 +356,7 @@ promote() {
 redeploy_tenant() {
   # args: <slug> — handle the 403→SSM-refresh→retry pattern
   local slug="$1"
+  validate_slug "$slug"
   log "  redeploy: $slug"
   if [[ "$DRY_RUN" == "true" ]]; then
     log "    [dry-run] would POST /redeploy slug=$slug"
@@ -372,6 +390,7 @@ redeploy_tenant() {
 
 verify_tenant() {
   local slug="$1"
+  validate_slug "$slug"
   log "  verify: $slug"
   if [[ "$DRY_RUN" == "true" ]]; then
     log "    [dry-run] would curl /buildinfo + /health"
@@ -398,6 +417,7 @@ rollback() {
   rm -f "$mfile"
   IFS=',' read -ra slugs <<<"$TENANTS"
   for slug in "${slugs[@]}"; do
+    validate_slug "$slug"
     redeploy_tenant "$slug" || err "  rollback redeploy failed for $slug"
   done
   log "rollback: complete"
@@ -408,6 +428,13 @@ rollback() {
 # ─────────────────────────────────────────────────────────────────────────────
 
 main() {
+  # OFFSEC-006: validate slugs before any network I/O.
+  IFS=',' read -ra _slugs <<<"$TENANTS"
+  for _slug in "${_slugs[@]}"; do
+    validate_slug "$_slug"
+  done
+  unset _slugs _slug
+
   preflight || return 1
   snapshot_dest_tag || return 2
   promote || return 2
@@ -415,8 +442,15 @@ main() {
   local promote_rc=0
   IFS=',' read -ra slugs <<<"$TENANTS"
   for slug in "${slugs[@]}"; do
-    redeploy_tenant "$slug" || promote_rc=1
-    [[ $promote_rc -eq 0 ]] && { verify_tenant "$slug" || promote_rc=1; }
+    validate_slug "$slug"
+    if ! redeploy_tenant "$slug"; then
+      promote_rc=1
+    fi
+    if [[ $promote_rc -eq 0 ]]; then
+      if ! verify_tenant "$slug"; then
+        promote_rc=1
+      fi
+    fi
     [[ $promote_rc -ne 0 ]] && break
   done
 

scripts/test-promote-tenant-image.sh

@@ -267,7 +267,51 @@ else
   printf '  ✗ unknown-flag should fail (got %s)\n' "$rc"
 fi
 
-printf '\n== Test 9: ROLLBACK_TAG follows YYYYMMDD via NOW_OVERRIDE_DATE ==\n'
+printf '\n== Test 9: slug validation — invalid slugs rejected with exit 64 (OFFSEC-006) ==\n'
+# Attack vectors: SSRF via ? (curl query separator), subdomain takeover via @,
+# path traversal via /, shell metacharacters.  Use a newline-delimited temp file
+# so slugs containing spaces are NOT split by shell word-splitting.
+_invalid_tmp=$(mktemp)
+cat > "$_invalid_tmp" <<'INVALID_EOF'
+a?url=https://evil.com
+a&url=https://evil.com
+a@evil.com
+a/b
+a\b
+a b
+chloe-dong?url=http://evil.com
+evil.com@legitimate
+INVALID_EOF
+while IFS= read -r attack || [[ -n "$attack" ]]; do
+  set +e
+  out=$("$SCRIPT" --source-tag x --dest-tag y --tenants "$attack" 2>&1); rc=$?
+  set -e
+  if [[ $rc -eq 64 ]] && printf '%s' "$out" | grep -q 'invalid slug'; then
+    PASS=$((PASS + 1)); printf '  ✓ slug rejected: %s\n' "$(printf '%q' "$attack")"
+  else
+    FAIL=$((FAIL + 1)); FAIL_NAMES+=("slug-reject:$attack")
+    printf '  ✗ slug should be rejected: %s — got exit %s\n' "$(printf '%q' "$attack")" "$rc"
+  fi
+done < "$_invalid_tmp"
+rm -f "$_invalid_tmp"
+
+printf '\n== Test 10: slug validation — valid slugs pass through ==\n'
+valid_slugs='chloe-dong hongming ab a abc123 my-tenant-42'
+for slug in $valid_slugs; do
+  set +e
+  out=$("$SCRIPT" --source-tag x --dest-tag y --tenants "$slug" --mock-dir /nonexistent 2>&1); rc=$?
+  set -e
+  # valid slugs: script should fail at preflight (no such mock dir / no real infra),
+  # but NOT at slug validation (exit 64). So we check exit != 64.
+  if [[ $rc -ne 64 ]]; then
+    PASS=$((PASS + 1)); printf '  ✓ valid slug accepted: %s\n' "$slug"
+  else
+    FAIL=$((FAIL + 1)); FAIL_NAMES+=("slug-accept:$slug")
+    printf '  ✗ valid slug rejected: %s (should have passed slug check)\n' "$slug"
+  fi
+done
+
+printf '\n== Test 11: ROLLBACK_TAG follows YYYYMMDD via NOW_OVERRIDE_DATE ==\n'
 m=$(mkmock)
 mock_set "$m" aws_ecr_get_image       '{}' 0
 mock_set "$m" aws_ecr_describe_image  '' 1
@@ -289,7 +333,7 @@ fi
 assert_calls_contain "rollback tag uses NOW_OVERRIDE_DATE (20260603)" "$m" 'aws_ecr_put_image b-prev-20260603'
 rm -rf "$m"
 
-printf '\n== Test 10: empty source manifest fails preflight ==\n'
+printf '\n== Test 12: empty source manifest fails preflight ==\n'
 m=$(mkmock)
 mock_set "$m" aws_ecr_get_image '' 0   # rc=0 but empty body (the "None" case)
 out=$(run_script "$m")
@@ -297,7 +341,7 @@ assert_exit "empty source manifest fails preflight" "$out" 1
 assert_contains "empty manifest message" "$out" 'returned empty manifest'
 rm -rf "$m"
 
-printf '\n== Test 11: tenant_buildinfo failure during verify → rollback ==\n'
+printf '\n== Test 13: tenant_buildinfo failure during verify → rollback ==\n'
 m=$(mkmock)
 mock_set "$m" aws_ecr_get_image          '{"manifests":[]}' 0
 mock_set "$m" aws_ecr_describe_image     '' 1
@@ -311,7 +355,7 @@ assert_contains "logs buildinfo failure" "$out" '/buildinfo failed for chloe-don
 assert_contains "rollback fired after verify fail" "$out" 'ROLLBACK:'
 rm -rf "$m"
 
-printf '\n== Test 12: ssm_refresh_ecr_auth JSON escaping (CWE-78 / OFFSEC-001) ==\n'
+printf '\n== Test 14: ssm_refresh_ecr_auth JSON escaping (CWE-78 / OFFSEC-001) ==\n'
 # Verify the python3 snippet in ssm_refresh_ecr_auth produces valid JSON and
 # correctly escapes shell-injection characters in region + account ID fields.
 # The fix replaces unquoted shell-printf interpolation with json.dumps.

tests/e2e/test_peer_visibility_mcp_staging.sh

@@ -0,0 +1,376 @@
+#!/usr/bin/env bash
+# Staging E2E — fresh-provision peer-visibility gate via the LITERAL MCP path.
+#
+# WHY THIS EXISTS
+# ---------------
+# Hermes and OpenClaw were repeatedly reported "fleet-verified / cascade-
+# complete" because the *proxy* signals were green:
+#   - registry-registration + heartbeat (Hermes), and
+#   - model round-trip 200 (OpenClaw).
+# But a freshly-provisioned workspace, asked on canvas "can you see your
+# peers", actually FAILS:
+#   - Hermes: 401 on the molecule MCP `list_peers` call,
+#   - OpenClaw: falls back to native `sessions_list`, sees no platform peers.
+# Tasks #142/#159 were even marked "completed" under this same proxy flaw.
+#
+# This script codifies the LITERAL user-facing path so it can never silently
+# regress: it provisions a brand-new throwaway org + sibling workspaces via
+# the real control-plane provisioning path, then for each runtime that should
+# have platform peer-visibility it drives the EXACT MCP call the canvas agent
+# makes — `POST /workspaces/:id/mcp` JSON-RPC tools/call name=list_peers,
+# authenticated by that workspace's own bearer token through the real
+# WorkspaceAuth + MCPRateLimiter middleware chain. It then asserts:
+#   (1) HTTP 200,
+#   (2) JSON-RPC `result` present (NOT an `error` object — a -32000
+#       "tool call failed" or a 401 from WorkspaceAuth fails here),
+#   (3) the returned peer set CONTAINS the other provisioned sibling
+#       workspace IDs — not an empty list, not a native-sessions fallback.
+#
+# This is NOT a proxy. It does not look at a registry row, /health, the
+# heartbeat table, or `GET /registry/:id/peers`. It drives the byte-for-byte
+# JSON-RPC envelope that mcp_molecule_list_peers issues from a real agent.
+#
+# It is written to FAIL on today's broken Hermes/OpenClaw behavior and go
+# green only when the in-flight root-cause fixes (Hermes-401, OpenClaw MCP
+# wiring) actually land. That is the point: it is the objective proof gate.
+#
+# AUTH MODEL (mirrors tests/e2e/test_staging_full_saas.sh)
+# --------------------------------------------------------
+#   Single MOLECULE_ADMIN_TOKEN (= CP_ADMIN_API_TOKEN on Railway staging)
+#   drives: POST /cp/admin/orgs (provision), GET
+#   /cp/admin/orgs/:slug/admin-token (per-tenant token), DELETE
+#   /cp/admin/tenants/:slug (teardown). The per-tenant admin token drives
+#   tenant workspace creation; each workspace's OWN auth_token (returned by
+#   POST /workspaces) drives its MCP call.
+#
+# Required env:
+#   MOLECULE_ADMIN_TOKEN   CP admin bearer — Railway staging CP_ADMIN_API_TOKEN
+# Optional env:
+#   MOLECULE_CP_URL        default https://staging-api.moleculesai.app
+#   E2E_RUN_ID             slug suffix; CI passes ${GITHUB_RUN_ID}
+#   PV_RUNTIMES            space list; default "hermes openclaw claude-code"
+#   E2E_PROVISION_TIMEOUT_SECS  default 1800 (hermes/openclaw cold EC2 budget)
+#   E2E_MINIMAX_API_KEY / E2E_ANTHROPIC_API_KEY / E2E_OPENAI_API_KEY
+#                          LLM provider key injected so the runtime can boot
+#   E2E_KEEP_ORG           1 → skip teardown (local debugging only)
+#
+# Exit codes:
+#   0  every runtime saw its peers via the literal MCP call
+#   1  generic failure
+#   2  missing required env
+#   3  provisioning timed out
+#   4  teardown left orphan resources
+#   10 peer-visibility regression reproduced (the gate firing as designed)
+
+set -uo pipefail
+
+CP_URL="${MOLECULE_CP_URL:-https://staging-api.moleculesai.app}"
+ADMIN_TOKEN="${MOLECULE_ADMIN_TOKEN:?MOLECULE_ADMIN_TOKEN required — Railway staging CP_ADMIN_API_TOKEN}"
+RUN_ID_SUFFIX="${E2E_RUN_ID:-$(date +%H%M%S)-$$}"
+PV_RUNTIMES="${PV_RUNTIMES:-hermes openclaw claude-code}"
+PROVISION_TIMEOUT_SECS="${E2E_PROVISION_TIMEOUT_SECS:-1800}"
+
+# Slug MUST start with 'e2e-' so the sweep-stale-e2e-orgs safety net
+# (EPHEMERAL_PREFIXES) catches any leak this run fails to tear down.
+SLUG="e2e-pv-$(date +%Y%m%d)-${RUN_ID_SUFFIX}"
+SLUG=$(echo "$SLUG" | tr '[:upper:]' '[:lower:]' | tr -cd 'a-z0-9-' | head -c 32)
+
+ORG_ID=""
+TENANT_URL=""
+TENANT_TOKEN=""
+
+log()  { echo "[$(date +%H:%M:%S)] $*"; }
+fail() { echo "[$(date +%H:%M:%S)] ❌ $*" >&2; exit 1; }
+ok()   { echo "[$(date +%H:%M:%S)] ✅ $*"; }
+
+admin_call() {
+  local method="$1" path="$2"; shift 2
+  curl -sS -X "$method" "$CP_URL$path" \
+    -H "Authorization: Bearer $ADMIN_TOKEN" \
+    -H "Content-Type: application/json" "$@"
+}
+tenant_call() {
+  local method="$1" path="$2"; shift 2
+  curl -sS -X "$method" "$TENANT_URL$path" \
+    -H "Authorization: Bearer $TENANT_TOKEN" \
+    -H "X-Molecule-Org-Id: $ORG_ID" \
+    -H "Content-Type: application/json" "$@"
+}
+
+# ─── Scoped teardown ───────────────────────────────────────────────────
+# Deletes ONLY the org this run created (DELETE /cp/admin/tenants/$SLUG
+# with the {"confirm":$SLUG} fat-finger guard). Never a cluster-wide
+# sweep — honors feedback_cleanup_after_each_test and
+# feedback_never_run_cluster_cleanup_tests_on_live_platform. The
+# workflow's always() step + sweep-stale-e2e-orgs are the outer nets.
+teardown() {
+  local rc=$?
+  set +e
+  if [ "${E2E_KEEP_ORG:-0}" = "1" ]; then
+    echo ""
+    log "[teardown] E2E_KEEP_ORG=1 — leaving $SLUG for debugging (REMEMBER TO DELETE)"
+    exit $rc
+  fi
+  echo ""
+  log "[teardown] DELETE /cp/admin/tenants/$SLUG (scoped to this run only)"
+  admin_call DELETE "/cp/admin/tenants/$SLUG" --max-time 120 \
+    -d "{\"confirm\":\"$SLUG\"}" >/dev/null 2>&1
+  for j in $(seq 1 24); do
+    LIST=$(admin_call GET "/cp/admin/orgs?limit=500" 2>/dev/null)
+    LEAK=$(echo "$LIST" | python3 -c "
+import sys, json
+try: d = json.load(sys.stdin)
+except Exception: print(1); sys.exit(0)
+orgs = d if isinstance(d, list) else d.get('orgs', [])
+print(sum(1 for o in orgs if o.get('slug') == '$SLUG' and o.get('instance_status') not in ('purged',) and o.get('status') != 'purged'))
+" 2>/dev/null || echo 1)
+    if [ "$LEAK" = "0" ]; then
+      log "[teardown] ✓ $SLUG purged (after ${j}x5s)"
+      exit $rc
+    fi
+    sleep 5
+  done
+  echo "::warning::[teardown] $SLUG still present after 120s — sweep-stale-e2e-orgs will catch it within MAX_AGE_MINUTES" >&2
+  [ $rc -eq 0 ] && rc=4
+  exit $rc
+}
+trap teardown EXIT INT TERM
+
+# ─── 1. Provision the throwaway org ────────────────────────────────────
+log "1/6 POST /cp/admin/orgs — slug=$SLUG"
+CREATE=$(admin_call POST /cp/admin/orgs \
+  -d "{\"slug\":\"$SLUG\",\"name\":\"E2E peer-visibility $SLUG\",\"owner_user_id\":\"e2e-runner:$SLUG\"}")
+ORG_ID=$(echo "$CREATE" | python3 -c "import sys,json; print(json.load(sys.stdin).get('id',''))" 2>/dev/null)
+[ -n "$ORG_ID" ] || fail "org creation failed: $(echo "$CREATE" | head -c 300)"
+log "    ORG_ID=$ORG_ID"
+
+# ─── 2. Wait for tenant EC2 + DNS ──────────────────────────────────────
+log "2/6 waiting for tenant instance_status=running (cold EC2 + cloudflared)..."
+DEADLINE=$(( $(date +%s) + PROVISION_TIMEOUT_SECS ))
+while true; do
+  [ "$(date +%s)" -gt "$DEADLINE" ] && fail "tenant never came up within ${PROVISION_TIMEOUT_SECS}s"
+  STATUS=$(admin_call GET "/cp/admin/orgs?limit=500" 2>/dev/null | python3 -c "
+import sys, json
+try: d = json.load(sys.stdin)
+except Exception: sys.exit(0)
+orgs = d if isinstance(d, list) else d.get('orgs', [])
+for o in orgs:
+    if o.get('slug') == '$SLUG':
+        print(o.get('instance_status') or o.get('status') or 'unknown'); break
+" 2>/dev/null)
+  case "$STATUS" in running|online|ready) break ;; esac
+  sleep 10
+done
+log "    tenant status=$STATUS"
+
+# ─── 3. Per-tenant admin token + tenant URL ────────────────────────────
+log "3/6 fetching per-tenant admin token..."
+TT_RESP=$(admin_call GET "/cp/admin/orgs/$SLUG/admin-token")
+TENANT_TOKEN=$(echo "$TT_RESP" | python3 -c "import sys,json; print(json.load(sys.stdin).get('admin_token',''))" 2>/dev/null)
+[ -n "$TENANT_TOKEN" ] || fail "tenant token fetch failed: $(echo "$TT_RESP" | head -c 200)"
+
+CP_HOST=$(echo "$CP_URL" | sed -E 's#^https?://##; s#/.*$##')
+case "$CP_HOST" in
+  api.*)         DERIVED_DOMAIN="${CP_HOST#api.}" ;;
+  staging-api.*) DERIVED_DOMAIN="staging.${CP_HOST#staging-api.}" ;;
+  *)             DERIVED_DOMAIN="$CP_HOST" ;;
+esac
+TENANT_URL="https://${SLUG}.${DERIVED_DOMAIN}"
+log "    tenant url: $TENANT_URL"
+
+log "3b. waiting for tenant /health (TLS/DNS, up to 10min)..."
+for i in $(seq 1 120); do
+  curl -fsS "$TENANT_URL/health" -m 5 -k >/dev/null 2>&1 && { log "    /health ok (attempt $i)"; break; }
+  sleep 5
+done
+
+# ─── 4. Provision the parent + one sibling per runtime under test ──────
+# Inject the LLM provider key so each runtime can authenticate at boot.
+# Priority: MiniMax → direct-Anthropic → OpenAI (mirrors
+# test_staging_full_saas.sh's secrets-injection chain).
+SECRETS_JSON='{}'
+if [ -n "${E2E_MINIMAX_API_KEY:-}" ]; then
+  SECRETS_JSON=$(python3 -c "import json,os;k=os.environ['E2E_MINIMAX_API_KEY'];print(json.dumps({'ANTHROPIC_BASE_URL':'https://api.minimax.io/anthropic','ANTHROPIC_AUTH_TOKEN':k,'MINIMAX_API_KEY':k}))")
+elif [ -n "${E2E_ANTHROPIC_API_KEY:-}" ]; then
+  SECRETS_JSON=$(python3 -c "import json,os;k=os.environ['E2E_ANTHROPIC_API_KEY'];print(json.dumps({'ANTHROPIC_API_KEY':k}))")
+elif [ -n "${E2E_OPENAI_API_KEY:-}" ]; then
+  SECRETS_JSON=$(python3 -c "import json,os;k=os.environ['E2E_OPENAI_API_KEY'];print(json.dumps({'OPENAI_API_KEY':k,'OPENAI_BASE_URL':'https://api.openai.com/v1','MODEL_PROVIDER':'openai:gpt-4o','HERMES_INFERENCE_PROVIDER':'custom','HERMES_CUSTOM_BASE_URL':'https://api.openai.com/v1','HERMES_CUSTOM_API_KEY':k,'HERMES_CUSTOM_API_MODE':'chat_completions'}))")
+fi
+
+log "4/6 provisioning parent (claude-code) + one sibling per runtime under test..."
+P_RESP=$(tenant_call POST /workspaces \
+  -d "{\"name\":\"pv-parent\",\"runtime\":\"claude-code\",\"tier\":3,\"secrets\":$SECRETS_JSON}")
+PARENT_ID=$(echo "$P_RESP" | python3 -c "import sys,json; print(json.load(sys.stdin).get('id',''))" 2>/dev/null)
+[ -n "$PARENT_ID" ] || fail "parent create failed: $(echo "$P_RESP" | head -c 300)"
+log "    PARENT_ID=$PARENT_ID"
+
+# WS_IDS[runtime]=id ; WS_TOKENS[runtime]=auth_token (the MCP bearer)
+declare -A WS_IDS WS_TOKENS
+ALL_WS_IDS="$PARENT_ID"
+for rt in $PV_RUNTIMES; do
+  R=$(tenant_call POST /workspaces \
+    -d "{\"name\":\"pv-$rt\",\"runtime\":\"$rt\",\"tier\":2,\"parent_id\":\"$PARENT_ID\",\"secrets\":$SECRETS_JSON}")
+  WID=$(echo "$R" | python3 -c "import sys,json; print(json.load(sys.stdin).get('id',''))" 2>/dev/null)
+  # auth_token is top-level for container runtimes; external-like nest it
+  # under connection.auth_token (verified vs staging response shape).
+  WTOK=$(echo "$R" | python3 -c "
+import sys, json
+try: d = json.load(sys.stdin)
+except Exception: print(''); sys.exit(0)
+print(d.get('auth_token') or d.get('connection', {}).get('auth_token') or '')
+" 2>/dev/null)
+  [ -n "$WID" ] || fail "$rt workspace create failed: $(echo "$R" | head -c 300)"
+  [ -n "$WTOK" ] || fail "$rt workspace did not return an auth_token — cannot drive its MCP call (resp: $(echo "$R" | head -c 300))"
+  WS_IDS[$rt]="$WID"
+  WS_TOKENS[$rt]="$WTOK"
+  ALL_WS_IDS="$ALL_WS_IDS $WID"
+  log "    $rt → $WID"
+done
+
+# ─── 5. Wait for every sibling online ──────────────────────────────────
+log "5/6 waiting for all workspaces status=online (up to ${PROVISION_TIMEOUT_SECS}s — cold boot)..."
+WS_DEADLINE=$(( $(date +%s) + PROVISION_TIMEOUT_SECS ))
+for rt in $PV_RUNTIMES; do
+  wid="${WS_IDS[$rt]}"
+  LAST=""
+  while true; do
+    [ "$(date +%s)" -gt "$WS_DEADLINE" ] && fail "$rt ($wid) never reached online (last=$LAST)"
+    S=$(tenant_call GET "/workspaces/$wid" 2>/dev/null | python3 -c "
+import sys, json
+try: d = json.load(sys.stdin)
+except Exception: sys.exit(0)
+w = d.get('workspace') if isinstance(d.get('workspace'), dict) else d
+print(w.get('status') or '')
+" 2>/dev/null)
+    [ "$S" != "$LAST" ] && { log "    $rt → $S"; LAST="$S"; }
+    case "$S" in
+      online) break ;;
+      failed) sleep 10 ;;   # transient: bootstrap-watcher 5-min deadline, heartbeat recovers
+      *)      sleep 10 ;;
+    esac
+  done
+  ok "    $rt online"
+done
+
+# ─── 6. THE GATE — literal mcp_molecule_list_peers via POST /:id/mcp ────
+# This is the byte-for-byte user-facing call. NOT GET /registry/:id/peers,
+# NOT /health, NOT the heartbeat table. JSON-RPC 2.0 tools/call,
+# name=list_peers, authenticated by the workspace's OWN bearer token
+# through WorkspaceAuth + MCPRateLimiter.
+log "6/6 driving the LITERAL list_peers MCP call per runtime..."
+echo ""
+RPC_BODY='{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"list_peers","arguments":{}}}'
+REGRESSED=0
+declare -A VERDICT
+
+for rt in $PV_RUNTIMES; do
+  wid="${WS_IDS[$rt]}"
+  wtok="${WS_TOKENS[$rt]}"
+  # The expected peer set = every OTHER provisioned workspace (parent +
+  # the sibling runtimes), excluding the caller itself.
+  EXPECT_IDS=$(echo "$ALL_WS_IDS" | tr ' ' '\n' | grep -v "^${wid}$" | grep -v '^$')
+
+  set +e
+  RESP=$(curl -sS -X POST "$TENANT_URL/workspaces/$wid/mcp" \
+    -H "Authorization: Bearer $wtok" \
+    -H "X-Molecule-Org-Id: $ORG_ID" \
+    -H "Content-Type: application/json" \
+    -d "$RPC_BODY" \
+    -o /tmp/pv_mcp_body.json -w "%{http_code}" 2>/dev/null)
+  set -e
+  HTTP_CODE="$RESP"
+  BODY=$(cat /tmp/pv_mcp_body.json 2>/dev/null || echo '')
+
+  echo "--- $rt (ws=$wid) ---"
+  echo "    HTTP $HTTP_CODE"
+  echo "    body: $(echo "$BODY" | head -c 600)"
+
+  # (1) HTTP 200 — a 401 (WorkspaceAuth reject, the Hermes symptom) fails here.
+  if [ "$HTTP_CODE" != "200" ]; then
+    echo "  ✗ $rt: list_peers MCP call returned HTTP $HTTP_CODE (expected 200)"
+    VERDICT[$rt]="FAIL(http=$HTTP_CODE)"
+    REGRESSED=1
+    continue
+  fi
+
+  # (2) JSON-RPC result present, not an error object.
+  PARSE=$(echo "$BODY" | python3 -c "
+import sys, json
+expect = set(filter(None, '''$EXPECT_IDS'''.split()))
+try:
+    d = json.load(sys.stdin)
+except Exception as e:
+    print('PARSE_ERROR:' + str(e)); sys.exit(0)
+if isinstance(d, dict) and d.get('error') is not None:
+    print('RPC_ERROR:' + json.dumps(d['error'])[:200]); sys.exit(0)
+res = d.get('result') if isinstance(d, dict) else None
+if res is None:
+    print('NO_RESULT'); sys.exit(0)
+# MCP tools/call result shape: {content:[{type:text,text:'<json or prose>'}]}
+text = ''
+if isinstance(res, dict):
+    for c in res.get('content', []):
+        if c.get('type') == 'text':
+            text += c.get('text', '')
+text_l = text.lower()
+# Native-sessions fallback signature (the OpenClaw symptom): the agent
+# answered from its own runtime session list, not the platform peer set.
+if 'sessions_list' in text_l or 'no platform peers' in text_l or 'native session' in text_l:
+    print('NATIVE_FALLBACK:' + text[:200]); sys.exit(0)
+# The expected sibling IDs must literally appear in the returned peer text.
+found = sorted(i for i in expect if i in text)
+missing = sorted(expect - set(found))
+if not expect:
+    print('NO_EXPECTED_PEERS_CONFIGURED'); sys.exit(0)
+if missing:
+    print('MISSING_PEERS:found=%d/%d missing=%s' % (len(found), len(expect), ','.join(m[:8] for m in missing)))
+    sys.exit(0)
+print('OK:found=%d/%d' % (len(found), len(expect)))
+" 2>/dev/null)
+
+  case "$PARSE" in
+    OK:*)
+      echo "  ✓ $rt: list_peers returned 200 and contains all expected peers ($PARSE)"
+      VERDICT[$rt]="OK"
+      ;;
+    NATIVE_FALLBACK:*)
+      echo "  ✗ $rt: list_peers fell back to NATIVE sessions — sees no platform peers ($PARSE)"
+      VERDICT[$rt]="FAIL(native-fallback)"
+      REGRESSED=1
+      ;;
+    RPC_ERROR:*|NO_RESULT|PARSE_ERROR:*)
+      echo "  ✗ $rt: list_peers MCP call did not return a usable result ($PARSE)"
+      VERDICT[$rt]="FAIL(rpc=$PARSE)"
+      REGRESSED=1
+      ;;
+    MISSING_PEERS:*)
+      echo "  ✗ $rt: list_peers returned 200 but peer set is wrong/empty ($PARSE)"
+      VERDICT[$rt]="FAIL(peers=$PARSE)"
+      REGRESSED=1
+      ;;
+    *)
+      echo "  ✗ $rt: unexpected verdict '$PARSE'"
+      VERDICT[$rt]="FAIL(unknown)"
+      REGRESSED=1
+      ;;
+  esac
+  echo ""
+done
+
+echo "=== SUMMARY — fresh-provision peer-visibility (literal MCP list_peers) ==="
+for rt in $PV_RUNTIMES; do
+  printf '  %-14s %s\n' "$rt" "${VERDICT[$rt]:-NO_RUN}"
+done
+echo ""
+
+if [ "$REGRESSED" -ne 0 ]; then
+  echo "✗ GATE FAILED — at least one runtime cannot see its peers via the"
+  echo "  literal mcp_molecule_list_peers call. This is the real user-facing"
+  echo "  failure the proxy signals (registry row / heartbeat / model 200)"
+  echo "  were hiding. Expected RED until the Hermes-401 + OpenClaw-MCP-wiring"
+  echo "  root-cause fixes land; goes green only when they actually do."
+  exit 10
+fi
+
+ok "GATE PASSED — every runtime under test sees its platform peers via the literal MCP call."
+exit 0

tests/e2e/test_workspace_abilities_e2e.sh

@@ -0,0 +1,296 @@
+#!/usr/bin/env bash
+# E2E test: workspace broadcast and talk-to-user platform abilities.
+#
+# What this proves:
+#   1. talk_to_user_enabled (default true) — POST /notify works out-of-the-box.
+#   2. PATCH /workspaces/:id/abilities { talk_to_user_enabled: false } disables
+#      delivery: /notify → 403 with error="talk_to_user_disabled" + delegate hint.
+#   3. Re-enabling talk_to_user_enabled restores delivery.
+#   4. broadcast_enabled (default false) — POST /broadcast → 403 when disabled.
+#   5. PATCH { broadcast_enabled: true } enables fan-out.
+#   6. POST /broadcast delivers to all non-sender, non-removed workspaces:
+#      - Returns {"status":"sent","delivered":N}
+#      - Receiver's activity log has a broadcast_receive entry with the message.
+#      - Sender's activity log has a broadcast_sent entry.
+#   7. The sender itself does NOT receive a broadcast_receive entry.
+#
+# Usage:  tests/e2e/test_workspace_abilities_e2e.sh
+# Prereqs: workspace-server on http://localhost:8080, MOLECULE_ENV != production
+
+set -euo pipefail
+
+source "$(dirname "$0")/_lib.sh"
+
+PASS=0
+FAIL=0
+SENDER_ID=""
+RECEIVER_ID=""
+
+cleanup() {
+  for wid in "$SENDER_ID" "$RECEIVER_ID"; do
+    if [ -n "$wid" ]; then
+      curl -s -X DELETE "$BASE/workspaces/$wid?confirm=true" > /dev/null || true
+    fi
+  done
+}
+trap cleanup EXIT INT TERM
+
+assert() {
+  local label="$1" actual="$2" expected="$3"
+  if [ "$actual" = "$expected" ]; then
+    echo "  PASS — $label"
+    PASS=$((PASS+1))
+  else
+    echo "  FAIL — $label"
+    echo "         expected: $expected"
+    echo "         actual:   $actual"
+    FAIL=$((FAIL+1))
+  fi
+}
+
+assert_contains() {
+  local label="$1" haystack="$2" needle="$3"
+  if echo "$haystack" | grep -qF "$needle"; then
+    echo "  PASS — $label"
+    PASS=$((PASS+1))
+  else
+    echo "  FAIL — $label"
+    echo "         needle:   $needle"
+    echo "         haystack: $haystack"
+    FAIL=$((FAIL+1))
+  fi
+}
+
+assert_not_contains() {
+  local label="$1" haystack="$2" needle="$3"
+  if ! echo "$haystack" | grep -qF "$needle"; then
+    echo "  PASS — $label"
+    PASS=$((PASS+1))
+  else
+    echo "  FAIL — $label (unexpected match)"
+    echo "         needle:   $needle"
+    echo "         haystack: $haystack"
+    FAIL=$((FAIL+1))
+  fi
+}
+
+# ── Pre-sweep: remove any stale leftover workspaces from a prior aborted run ──
+echo "=== Setup ==="
+for NAME in "Abilities Sender" "Abilities Receiver"; do
+  PRIOR=$(curl -s "$BASE/workspaces" | python3 -c "
+import json, sys
+try:
+    print(' '.join(w['id'] for w in json.load(sys.stdin) if w.get('name') == '$NAME'))
+except Exception:
+    pass
+")
+  for _wid in $PRIOR; do
+    echo "Sweeping leftover '$NAME' workspace: $_wid"
+    curl -s -X DELETE "$BASE/workspaces/$_wid?confirm=true" > /dev/null || true
+  done
+done
+
+R=$(curl -s -X POST "$BASE/workspaces" -H "Content-Type: application/json" \
+  -d '{"name":"Abilities Sender","tier":1}')
+SENDER_ID=$(echo "$R" | python3 -c 'import json,sys;print(json.load(sys.stdin)["id"])' 2>/dev/null || true)
+[ -n "$SENDER_ID" ] || { echo "Failed to create sender workspace: $R"; exit 1; }
+echo "Created sender   workspace: $SENDER_ID"
+
+R=$(curl -s -X POST "$BASE/workspaces" -H "Content-Type: application/json" \
+  -d '{"name":"Abilities Receiver","tier":1}')
+RECEIVER_ID=$(echo "$R" | python3 -c 'import json,sys;print(json.load(sys.stdin)["id"])' 2>/dev/null || true)
+[ -n "$RECEIVER_ID" ] || { echo "Failed to create receiver workspace: $R"; exit 1; }
+echo "Created receiver workspace: $RECEIVER_ID"
+
+# Mint workspace-scoped bearer tokens (test-only endpoint, disabled in prod).
+SENDER_TOKEN=$(e2e_mint_test_token "$SENDER_ID")
+[ -n "$SENDER_TOKEN" ] || { echo "Failed to mint sender token"; exit 1; }
+SENDER_AUTH="Authorization: Bearer $SENDER_TOKEN"
+
+# Admin token — any live workspace bearer satisfies AdminAuth in local dev.
+# In production-like envs, set MOLECULE_ADMIN_TOKEN.
+ADMIN_TOKEN="${MOLECULE_ADMIN_TOKEN:-$SENDER_TOKEN}"
+ADMIN_AUTH="Authorization: Bearer $ADMIN_TOKEN"
+
+# ─────────────────────────────────────────────────────────────────────────────
+echo ""
+echo "=== Part 1: talk_to_user ability ==="
+
+echo ""
+echo "--- 1a: /notify works with default talk_to_user_enabled=true ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$BASE/workspaces/$SENDER_ID/notify" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Hello from sender"}')
+assert "POST /notify returns 200 when talk_to_user_enabled=true (default)" "$CODE" "200"
+
+echo ""
+echo "--- 1b: Disable talk_to_user ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X PATCH "$BASE/workspaces/$SENDER_ID/abilities" \
+  -H "Content-Type: application/json" -H "$ADMIN_AUTH" \
+  -d '{"talk_to_user_enabled": false}')
+assert "PATCH /abilities talk_to_user_enabled=false returns 200" "$CODE" "200"
+
+# Verify the flag is reflected in the workspace GET response.
+WS=$(curl -s "$BASE/workspaces/$SENDER_ID" -H "$SENDER_AUTH")
+FLAG=$(echo "$WS" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("talk_to_user_enabled","MISSING"))')
+assert "GET /workspaces/:id reflects talk_to_user_enabled=false" "$FLAG" "False"
+
+echo ""
+echo "--- 1c: /notify blocked when talk_to_user disabled ---"
+BODY=$(curl -s -w "" -X POST "$BASE/workspaces/$SENDER_ID/notify" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Should be blocked"}')
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$BASE/workspaces/$SENDER_ID/notify" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Should be blocked"}')
+assert "POST /notify returns 403 when talk_to_user_enabled=false" "$CODE" "403"
+
+ERR=$(echo "$BODY" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("error",""))' 2>/dev/null || echo "")
+assert_contains "403 body contains talk_to_user_disabled error code" "$ERR" "talk_to_user_disabled"
+
+HINT=$(echo "$BODY" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("hint",""))' 2>/dev/null || echo "")
+assert_contains "403 body contains delegate_task hint" "$HINT" "delegate_task"
+
+echo ""
+echo "--- 1d: Re-enable talk_to_user and verify /notify works again ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X PATCH "$BASE/workspaces/$SENDER_ID/abilities" \
+  -H "Content-Type: application/json" -H "$ADMIN_AUTH" \
+  -d '{"talk_to_user_enabled": true}')
+assert "PATCH /abilities talk_to_user_enabled=true returns 200" "$CODE" "200"
+
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$BASE/workspaces/$SENDER_ID/notify" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Re-enabled, should work"}')
+assert "POST /notify returns 200 after re-enabling talk_to_user" "$CODE" "200"
+
+# ─────────────────────────────────────────────────────────────────────────────
+echo ""
+echo "=== Part 2: broadcast ability ==="
+
+echo ""
+echo "--- 2a: Broadcast blocked by default (broadcast_enabled=false) ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$BASE/workspaces/$SENDER_ID/broadcast" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Should be blocked"}')
+assert "POST /broadcast returns 403 when broadcast_enabled=false (default)" "$CODE" "403"
+
+echo ""
+echo "--- 2b: Enable broadcast ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X PATCH "$BASE/workspaces/$SENDER_ID/abilities" \
+  -H "Content-Type: application/json" -H "$ADMIN_AUTH" \
+  -d '{"broadcast_enabled": true}')
+assert "PATCH /abilities broadcast_enabled=true returns 200" "$CODE" "200"
+
+WS=$(curl -s "$BASE/workspaces/$SENDER_ID" -H "$SENDER_AUTH")
+FLAG=$(echo "$WS" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("broadcast_enabled","MISSING"))')
+assert "GET /workspaces/:id reflects broadcast_enabled=true" "$FLAG" "True"
+
+echo ""
+echo "--- 2c: Successful broadcast fan-out ---"
+BCAST=$(curl -s -X POST "$BASE/workspaces/$SENDER_ID/broadcast" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":"Org-wide notice: scheduled maintenance in 5 minutes."}')
+BSTATUS=$(echo "$BCAST" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("status",""))' 2>/dev/null || echo "")
+BDELIVERED=$(echo "$BCAST" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("delivered","-1"))' 2>/dev/null || echo "-1")
+assert "POST /broadcast returns status=sent" "$BSTATUS" "sent"
+
+# delivered count must be >= 1 (the receiver workspace).
+echo "  INFO — broadcast delivered=$BDELIVERED"
+if python3 -c "import sys; sys.exit(0 if int('$BDELIVERED') >= 1 else 1)" 2>/dev/null; then
+  echo "  PASS — delivered count >= 1"
+  PASS=$((PASS+1))
+else
+  echo "  FAIL — expected delivered >= 1, got $BDELIVERED"
+  FAIL=$((FAIL+1))
+fi
+
+echo ""
+echo "--- 2d: Receiver activity log has broadcast_receive entry ---"
+RECEIVER_TOKEN=$(e2e_mint_test_token "$RECEIVER_ID")
+[ -n "$RECEIVER_TOKEN" ] || { echo "Failed to mint receiver token"; exit 1; }
+RECEIVER_AUTH="Authorization: Bearer $RECEIVER_TOKEN"
+
+ACT=$(curl -s -H "$RECEIVER_AUTH" "$BASE/workspaces/$RECEIVER_ID/activity?source=agent&limit=20")
+ROW=$(echo "$ACT" | python3 -c '
+import json, sys
+rows = json.load(sys.stdin) or []
+for r in rows:
+    if r.get("activity_type") == "broadcast_receive":
+        print(json.dumps(r))
+        break
+')
+[ -n "$ROW" ] || {
+  echo "  FAIL — could not find broadcast_receive row in receiver activity"
+  FAIL=$((FAIL+1))
+}
+
+if [ -n "$ROW" ]; then
+  # Message is stored in summary field.
+  MSG=$(echo "$ROW" | python3 -c 'import json,sys;r=json.load(sys.stdin);print(r.get("summary",""))')
+  assert_contains "broadcast_receive row summary has original message" "$MSG" "scheduled maintenance"
+  # Sender ID is stored in source_id field.
+  SRC=$(echo "$ROW" | python3 -c 'import json,sys;r=json.load(sys.stdin);print(r.get("source_id",""))')
+  assert "broadcast_receive row source_id is sender workspace" "$SRC" "$SENDER_ID"
+fi
+
+echo ""
+echo "--- 2e: Sender activity log has broadcast_sent entry ---"
+ACT_SENDER=$(curl -s -H "$SENDER_AUTH" "$BASE/workspaces/$SENDER_ID/activity?limit=20")
+SENT_ROW=$(echo "$ACT_SENDER" | python3 -c '
+import json, sys
+rows = json.load(sys.stdin) or []
+for r in rows:
+    if r.get("activity_type") == "broadcast_sent":
+        print(json.dumps(r))
+        break
+')
+[ -n "$SENT_ROW" ] || {
+  echo "  FAIL — could not find broadcast_sent row in sender activity"
+  FAIL=$((FAIL+1))
+}
+
+if [ -n "$SENT_ROW" ]; then
+  # Delivered count is baked into the summary field (no response_body for sender row).
+  SUMMARY=$(echo "$SENT_ROW" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("summary",""))')
+  assert_contains "broadcast_sent summary mentions workspace count" "$SUMMARY" "workspace"
+fi
+
+echo ""
+echo "--- 2f: Sender does NOT receive a broadcast_receive entry ---"
+SELF_RECV=$(echo "$ACT_SENDER" | python3 -c '
+import json, sys
+rows = json.load(sys.stdin) or []
+for r in rows:
+    if r.get("activity_type") == "broadcast_receive":
+        print("found")
+        break
+')
+assert_not_contains "sender has no broadcast_receive in own activity log" "${SELF_RECV:-}" "found"
+
+# ─────────────────────────────────────────────────────────────────────────────
+echo ""
+echo "--- 2g: Empty message is rejected ---"
+CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST "$BASE/workspaces/$SENDER_ID/broadcast" \
+  -H "Content-Type: application/json" -H "$SENDER_AUTH" \
+  -d '{"message":""}')
+assert "POST /broadcast with empty message returns 400" "$CODE" "400"
+
+echo ""
+echo "--- 2h: Partial PATCH does not clobber other flags ---"
+# Set talk_to_user=false, then patch only broadcast — talk_to_user must stay false.
+curl -s -o /dev/null -X PATCH "$BASE/workspaces/$SENDER_ID/abilities" \
+  -H "Content-Type: application/json" -H "$ADMIN_AUTH" \
+  -d '{"talk_to_user_enabled": false}'
+curl -s -o /dev/null -X PATCH "$BASE/workspaces/$SENDER_ID/abilities" \
+  -H "Content-Type: application/json" -H "$ADMIN_AUTH" \
+  -d '{"broadcast_enabled": false}'
+WS=$(curl -s "$BASE/workspaces/$SENDER_ID" -H "$SENDER_AUTH")
+TUF=$(echo "$WS" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("talk_to_user_enabled","MISSING"))')
+BEF=$(echo "$WS" | python3 -c 'import json,sys;print(json.load(sys.stdin).get("broadcast_enabled","MISSING"))')
+assert "partial PATCH preserves talk_to_user_enabled=false" "$TUF" "False"
+assert "partial PATCH sets broadcast_enabled=false" "$BEF" "False"
+
+# ─────────────────────────────────────────────────────────────────────────────
+echo ""
+echo "=== Results: $PASS passed, $FAIL failed ==="
+[ "$FAIL" -eq 0 ]

tests/test_lint_workflow_yaml.py

@@ -22,6 +22,7 @@ Cross-links:
 """
 from __future__ import annotations
 
+import re
 import subprocess
 import sys
 import textwrap
@@ -542,3 +543,153 @@ def test_rule9_prod_manual_deploy_allows_rollback_control(tmp_path):
     _write(tmp_path, "ok.yml", PROD_ROLLBACK_OK)
     r = _run_lint(tmp_path)
     assert r.returncode == 0, f"stdout={r.stdout}\nstderr={r.stderr}"
+
+
+# ---------------------------------------------------------------------------
+# Rule 10 — docker info piped to head under pipefail
+# ---------------------------------------------------------------------------
+
+DOCKER_INFO_HEAD_BAD = """
+    name: docker-info-head-bad
+    on: [push]
+    jobs:
+      build:
+        runs-on: ubuntu-latest
+        steps:
+          - run: |
+              set -euo pipefail
+              docker info 2>&1 | head -5 || exit 1
+"""
+
+DOCKER_INFO_CAPTURE_OK = """
+    name: docker-info-capture-ok
+    on: [push]
+    jobs:
+      build:
+        runs-on: ubuntu-latest
+        steps:
+          - run: |
+              set -euo pipefail
+              docker_info="$(docker info 2>&1)" || exit 1
+              printf '%s\\n' "${docker_info}" | sed -n '1,5p'
+"""
+
+DOCKER_INFO_SEPARATE_STEP_OK = """
+    name: docker-info-separate-step-ok
+    on: [push]
+    jobs:
+      build:
+        runs-on: ubuntu-latest
+        steps:
+          - run: |
+              set -euo pipefail
+              echo setup
+          - run: |
+              docker info 2>&1 | head -5 || true
+"""
+
+
+def test_rule10_docker_info_head_under_pipefail_detects_violation(tmp_path):
+    _write(tmp_path, "bad.yml", DOCKER_INFO_HEAD_BAD)
+    r = _run_lint(tmp_path)
+    assert r.returncode == 1
+    assert "docker info" in r.stdout.lower()
+    assert "pipefail" in r.stdout.lower()
+
+
+def test_rule10_docker_info_capture_passes(tmp_path):
+    _write(tmp_path, "ok.yml", DOCKER_INFO_CAPTURE_OK)
+    r = _run_lint(tmp_path)
+    assert r.returncode == 0, f"stdout={r.stdout}\nstderr={r.stderr}"
+
+
+def test_rule10_docker_info_head_in_separate_step_without_pipefail_passes(tmp_path):
+    _write(tmp_path, "ok.yml", DOCKER_INFO_SEPARATE_STEP_OK)
+    r = _run_lint(tmp_path)
+    assert r.returncode == 0, f"stdout={r.stdout}\nstderr={r.stderr}"
+
+
+# ---------------------------------------------------------------------------
+# CI change detector fanout — workflow-only PRs keep required contexts without
+# running Go/Canvas/Python/shellcheck heavy steps.
+# ---------------------------------------------------------------------------
+
+CI_WORKFLOW = REPO_ROOT / ".gitea" / "workflows" / "ci.yml"
+CI_SURFACES = ("platform", "canvas", "python", "scripts")
+
+
+def _ci_change_patterns() -> dict[str, re.Pattern[str]]:
+    text = CI_WORKFLOW.read_text(encoding="utf-8")
+    patterns: dict[str, re.Pattern[str]] = {}
+    for surface, pattern in re.findall(
+        r'echo "(platform|canvas|python|scripts)=.*?grep -qE \'([^\']+)\'',
+        text,
+    ):
+        patterns[surface] = re.compile(pattern)
+    assert set(patterns) == set(CI_SURFACES)
+    return patterns
+
+
+def _classify_ci_change(*paths: str) -> dict[str, bool]:
+    patterns = _ci_change_patterns()
+    return {
+        surface: any(pattern.search(path) for path in paths)
+        for surface, pattern in patterns.items()
+    }
+
+
+def test_ci_change_detector_workflow_only_edits_do_not_trigger_heavy_surfaces():
+    assert _classify_ci_change(".gitea/workflows/ci.yml") == {
+        "platform": False,
+        "canvas": False,
+        "python": False,
+        "scripts": False,
+    }
+    assert _classify_ci_change(".github/workflows/ci.yml") == {
+        "platform": False,
+        "canvas": False,
+        "python": False,
+        "scripts": False,
+    }
+
+
+def test_ci_change_detector_narrow_surface_edits_only_trigger_their_surface():
+    assert _classify_ci_change("workspace-server/internal/handlers/foo.go") == {
+        "platform": True,
+        "canvas": False,
+        "python": False,
+        "scripts": False,
+    }
+    assert _classify_ci_change("canvas/app/page.tsx") == {
+        "platform": False,
+        "canvas": True,
+        "python": False,
+        "scripts": False,
+    }
+    assert _classify_ci_change("workspace/a2a_mcp_server.py") == {
+        "platform": False,
+        "canvas": False,
+        "python": True,
+        "scripts": False,
+    }
+    assert _classify_ci_change("tests/e2e/test_model_slug.sh") == {
+        "platform": False,
+        "canvas": False,
+        "python": False,
+        "scripts": True,
+    }
+
+
+def test_ci_change_detector_docs_and_meta_scripts_do_not_trigger_surfaces():
+    assert _classify_ci_change("README.md") == {
+        "platform": False,
+        "canvas": False,
+        "python": False,
+        "scripts": False,
+    }
+    assert _classify_ci_change(".gitea/scripts/lint-workflow-yaml.py") == {
+        "platform": False,
+        "canvas": False,
+        "python": False,
+        "scripts": False,
+    }

tests/test_status_reaper.py

@@ -495,7 +495,7 @@ def test_reap_required_check_pull_request_suffix_never_touched(sr_module, monkey
     }
     counters = sr_module.reap(workflow_map, combined, SHA, dry_run=False)
     assert counters["compensated"] == 0
-    assert counters["preserved_non_push_suffix"] == 1
+    assert counters["preserved_pr_without_push_success"] == 1
     assert calls == []
 
 
@@ -1009,3 +1009,64 @@ def test_reap_continues_on_per_sha_apierror(sr_module, monkeypatch, capsys):
     captured = capsys.readouterr()
     assert "::warning::" in captured.out or "::notice::" in captured.out
     assert SHA_A[:10] in captured.out
+
+
+def test_main_soft_skips_when_commit_listing_times_out(sr_module, monkeypatch, capsys):
+    """A transient outage while listing recent commits should not paint main red.
+
+    Per-SHA status read failures are already isolated inside `reap_branch`.
+    The real 2026-05-14 failure was earlier: `/commits?sha=main&limit=30`
+    timed out after all retries, aborting the tick. The next 5-minute tick can
+    retry safely, so `main()` should emit an observable warning and return 0.
+    """
+
+    monkeypatch.setattr(sr_module, "scan_workflows", lambda _: {"workflow-without-push": False})
+
+    def fake_list_recent_commit_shas(*args, **kwargs):
+        raise sr_module.ApiError(
+            "GET /repos/owner/repo/commits failed after 4 attempts: timed out"
+        )
+
+    monkeypatch.setattr(sr_module, "list_recent_commit_shas", fake_list_recent_commit_shas)
+    monkeypatch.setattr(sys, "argv", ["status-reaper.py"])
+
+    assert sr_module.main() == 0
+    captured = capsys.readouterr()
+    assert "::warning::status-reaper skipped this tick" in captured.out
+    assert '"skipped": true' in captured.out
+    assert '"skip_reason": "commit-list-api-error"' in captured.out
+
+
+def test_main_does_not_soft_skip_status_write_failures(sr_module, monkeypatch):
+    """Only commit-list read failures are soft-skipped.
+
+    A compensation write failure means the reaper could not repair a red
+    status. That must still fail the job loudly instead of being mislabeled as
+    a transient commit-list outage.
+    """
+
+    monkeypatch.setattr(sr_module, "scan_workflows", lambda _: {"workflow-without-push": False})
+    monkeypatch.setattr(sr_module, "list_recent_commit_shas", lambda *_args, **_kwargs: [SHA_A])
+    monkeypatch.setattr(
+        sr_module,
+        "get_combined_status",
+        lambda _sha: {
+            "state": "failure",
+            "statuses": [
+                {
+                    "context": "workflow-without-push / job (push)",
+                    "status": "failure",
+                    "description": "stranded class-O red",
+                }
+            ],
+        },
+    )
+
+    def fake_post_compensating_status(*args, **kwargs):
+        raise sr_module.ApiError("POST /statuses failed: 403")
+
+    monkeypatch.setattr(sr_module, "post_compensating_status", fake_post_compensating_status)
+    monkeypatch.setattr(sys, "argv", ["status-reaper.py"])
+
+    with pytest.raises(sr_module.ApiError, match="POST /statuses failed"):
+        sr_module.main()

workspace-server/go.mod

@@ -18,6 +18,7 @@ require (
 	github.com/opencontainers/image-spec v1.1.1
 	github.com/redis/go-redis/v9 v9.19.0
 	github.com/robfig/cron/v3 v3.0.1
+	github.com/stretchr/testify v1.11.1
 	go.moleculesai.app/plugin/gh-identity v0.0.0-20260509010445-788988195fce
 	golang.org/x/crypto v0.50.0
 	gopkg.in/yaml.v3 v3.0.1
@@ -33,6 +34,7 @@ require (
 	github.com/containerd/errdefs v1.0.0 // indirect
 	github.com/containerd/errdefs/pkg v0.3.0 // indirect
 	github.com/containerd/log v0.1.0 // indirect
+	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/distribution/reference v0.6.0 // indirect
 	github.com/docker/go-units v0.5.0 // indirect
 	github.com/felixge/httpsnoop v1.0.4 // indirect
@@ -58,6 +60,7 @@ require (
 	github.com/opencontainers/go-digest v1.0.0 // indirect
 	github.com/pelletier/go-toml/v2 v2.2.4 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
+	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/quic-go/qpack v0.6.0 // indirect
 	github.com/quic-go/quic-go v0.59.0 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect

workspace-server/internal/bundle/exporter_test.go

@@ -0,0 +1,261 @@
+package bundle
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+// ---------------------------------------------------------------------------
+// extractDescription
+// ---------------------------------------------------------------------------
+
+func TestExtractDescription_WithFrontmatter(t *testing.T) {
+	// YAML frontmatter is skipped; first non-comment, non-empty line after
+	// the closing `---` is the description.
+	content := `---
+title: My Workspace
+---
+# This is a comment
+This is the description line.
+Another line.`
+	got := extractDescription(content)
+	if got != "This is the description line." {
+		t.Errorf("got %q, want %q", got, "This is the description line.")
+	}
+}
+
+func TestExtractDescription_NoFrontmatter(t *testing.T) {
+	// No frontmatter: first non-comment, non-empty line is returned.
+	content := `# Copyright header
+My workspace description
+Another line.`
+	got := extractDescription(content)
+	if got != "My workspace description" {
+		t.Errorf("got %q, want %q", got, "My workspace description")
+	}
+}
+
+func TestExtractDescription_CommentOnly(t *testing.T) {
+	// All content is comments or empty → empty string.
+	content := `# comment only
+# another comment
+`
+	got := extractDescription(content)
+	if got != "" {
+		t.Errorf("got %q, want empty string", got)
+	}
+}
+
+func TestExtractDescription_EmptyInput(t *testing.T) {
+	got := extractDescription("")
+	if got != "" {
+		t.Errorf("got %q, want empty string", got)
+	}
+}
+
+func TestExtractDescription_UnclosedFrontmatter(t *testing.T) {
+	// With no closing `---`, inFrontmatter stays true after the opening
+	// delimiter, so all subsequent lines are skipped and "" is returned.
+	// This is the documented behaviour: without a closing delimiter,
+	// all lines are considered frontmatter.
+	content := `---
+title: No closing delimiter
+This is the description.`
+	got := extractDescription(content)
+	if got != "" {
+		t.Errorf("unclosed frontmatter: got %q, want empty string", got)
+	}
+}
+
+func TestExtractDescription_FrontmatterThenCommentThenContent(t *testing.T) {
+	content := `---
+tags: [test]
+---
+# internal comment
+Real description here.
+`
+	got := extractDescription(content)
+	if got != "Real description here." {
+		t.Errorf("got %q, want %q", got, "Real description here.")
+	}
+}
+
+func TestExtractDescription_BlankLinesSkipped(t *testing.T) {
+	// Empty lines (len=0) are skipped; whitespace-only lines (spaces) are NOT
+	// skipped because len(line)>0. First non-comment, non-empty line is returned.
+	content := "\n\n\n\nA. Description\nB. Should not be returned.\n"
+	got := extractDescription(content)
+	if got != "A. Description" {
+		t.Errorf("got %q, want %q", got, "A. Description")
+	}
+}
+
+// ---------------------------------------------------------------------------
+// splitLines
+// ---------------------------------------------------------------------------
+
+func TestSplitLines_Basic(t *testing.T) {
+	got := splitLines("a\nb\nc")
+	want := []string{"a", "b", "c"}
+	if len(got) != len(want) {
+		t.Fatalf("len=%d, want %d", len(got), len(want))
+	}
+	for i := range want {
+		if got[i] != want[i] {
+			t.Errorf("got[%d]=%q, want %q", i, got[i], want[i])
+		}
+	}
+}
+
+func TestSplitLines_TrailingNewline(t *testing.T) {
+	got := splitLines("line1\nline2\n")
+	want := []string{"line1", "line2"}
+	if len(got) != len(want) {
+		t.Errorf("trailing newline: got %v, want %v", got, want)
+	}
+}
+
+func TestSplitLines_NoNewline(t *testing.T) {
+	got := splitLines("no newline")
+	want := []string{"no newline"}
+	if len(got) != 1 || got[0] != want[0] {
+		t.Errorf("got %v, want %v", got, want)
+	}
+}
+
+func TestSplitLines_EmptyString(t *testing.T) {
+	got := splitLines("")
+	if len(got) != 0 {
+		t.Errorf("empty string: got %v, want []", got)
+	}
+}
+
+func TestSplitLines_OnlyNewlines(t *testing.T) {
+	got := splitLines("\n\n\n")
+	// Three consecutive '\n' characters → s[start:i] at each '\n' gives
+	// the empty string between newlines → 3 empty segments.
+	// (No trailing segment because start == len(s) at the end.)
+	if len(got) != 3 {
+		t.Errorf("only newlines: got %v (len=%d), want 3 empty strings", got, len(got))
+	}
+	for i, s := range got {
+		if s != "" {
+			t.Errorf("got[%d]=%q, want empty string", i, s)
+		}
+	}
+}
+
+func TestSplitLines_MultipleConsecutiveNewlines(t *testing.T) {
+	got := splitLines("a\n\n\nb")
+	// a\n\n\nb → ["a", "", "", "b"]
+	if len(got) != 4 {
+		t.Errorf("consecutive newlines: got %v (len=%d)", got, len(got))
+	}
+	if got[0] != "a" || got[3] != "b" {
+		t.Errorf("first/last: got %v, want [a, ..., b]", got)
+	}
+}
+
+// ---------------------------------------------------------------------------
+// findConfigDir
+// ---------------------------------------------------------------------------
+
+func TestFindConfigDir_NameMatch(t *testing.T) {
+	tmp := t.TempDir()
+
+	// Create two sub-dirs; only the one with matching name should be found.
+	mustMkdir(filepath.Join(tmp, "workspace-a"))
+	mustWrite(filepath.Join(tmp, "workspace-a", "config.yaml"),
+		"name: other-workspace\ntier: 1\n")
+
+	mustMkdir(filepath.Join(tmp, "workspace-b"))
+	mustWrite(filepath.Join(tmp, "workspace-b", "config.yaml"),
+		"name: target-workspace\nruntime: claude-code\n")
+
+	got := findConfigDir(tmp, "target-workspace")
+	want := filepath.Join(tmp, "workspace-b")
+	if got != want {
+		t.Errorf("got %q, want %q", got, want)
+	}
+}
+
+func TestFindConfigDir_NoMatch_UsesFallback(t *testing.T) {
+	tmp := t.TempDir()
+
+	mustMkdir(filepath.Join(tmp, "first"))
+	mustWrite(filepath.Join(tmp, "first", "config.yaml"), "name: workspace-a\n")
+
+	mustMkdir(filepath.Join(tmp, "second"))
+	mustWrite(filepath.Join(tmp, "second", "config.yaml"), "name: workspace-b\n")
+
+	// No exact name match → fallback to the first directory with a config.yaml.
+	got := findConfigDir(tmp, "nonexistent")
+	want := filepath.Join(tmp, "first")
+	if got != want {
+		t.Errorf("no match: got %q, want fallback %q", got, want)
+	}
+}
+
+func TestFindConfigDir_MissingDir(t *testing.T) {
+	got := findConfigDir("/nonexistent/path/for/findConfigDir", "any-name")
+	if got != "" {
+		t.Errorf("missing dir: got %q, want empty string", got)
+	}
+}
+
+func TestFindConfigDir_NoSubdirs(t *testing.T) {
+	tmp := t.TempDir()
+	// Empty directory → no matches, no fallback.
+	got := findConfigDir(tmp, "any")
+	if got != "" {
+		t.Errorf("empty dir: got %q, want empty string", got)
+	}
+}
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+func mustMkdir(path string) {
+	os.MkdirAll(path, 0o755)
+}
+
+func mustWrite(path, content string) {
+	os.WriteFile(path, []byte(content), 0o644)
+}
+
+// ---------------------------------------------------------------------------
+// findConfigDir
+// ---------------------------------------------------------------------------
+
+func TestFindConfigDir_SubdirWithoutConfig(t *testing.T) {
+	tmp := t.TempDir()
+	mustMkdir(filepath.Join(tmp, "empty-skill"))
+	// Sub-dir without config.yaml → skipped.
+	got := findConfigDir(tmp, "any")
+	if got != "" {
+		t.Errorf("no config.yaml: got %q, want empty string", got)
+	}
+}
+
+func TestFindConfigDir_FirstWithConfigIsFallback(t *testing.T) {
+	// When name doesn't match, fallback is the FIRST dir with config.yaml,
+	// not the last. Confirm ordering by creating three dirs.
+	tmp := t.TempDir()
+
+	mustMkdir(filepath.Join(tmp, "a"))
+	mustWrite(filepath.Join(tmp, "a", "config.yaml"), "name: alpha\n")
+
+	mustMkdir(filepath.Join(tmp, "b"))
+	mustWrite(filepath.Join(tmp, "b", "config.yaml"), "name: beta\n")
+
+	mustMkdir(filepath.Join(tmp, "c"))
+	mustWrite(filepath.Join(tmp, "c", "config.yaml"), "name: gamma\n")
+
+	got := findConfigDir(tmp, "nonexistent")
+	want := filepath.Join(tmp, "a") // first dir with config.yaml
+	if got != want {
+		t.Errorf("fallback order: got %q, want first-with-config %q", got, want)
+	}
+}

workspace-server/internal/bundle/importer_test.go

@@ -0,0 +1,317 @@
+package bundle
+
+import (
+	"testing"
+)
+
+func TestBuildBundleConfigFiles_EmptyBundle(t *testing.T) {
+	b := &Bundle{}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 0 {
+		t.Errorf("empty bundle: want 0 files, got %d", len(files))
+	}
+}
+
+func TestBuildBundleConfigFiles_SystemPromptOnly(t *testing.T) {
+	b := &Bundle{
+		SystemPrompt: "You are a helpful assistant.",
+	}
+	files := buildBundleConfigFiles(b)
+	if n := len(files); n != 1 {
+		t.Fatalf("system-prompt only: want 1 file, got %d", n)
+	}
+	if content, ok := files["system-prompt.md"]; !ok {
+		t.Fatal("missing system-prompt.md")
+	} else if string(content) != "You are a helpful assistant." {
+		t.Errorf("system-prompt content: got %q", string(content))
+	}
+}
+
+func TestBuildBundleConfigFiles_ConfigYamlOnly(t *testing.T) {
+	b := &Bundle{
+		Prompts: map[string]string{
+			"config.yaml": "runtime: langgraph\ntier: 2\n",
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	if n := len(files); n != 1 {
+		t.Fatalf("config.yaml only: want 1 file, got %d", n)
+	}
+	if content, ok := files["config.yaml"]; !ok {
+		t.Fatal("missing config.yaml")
+	} else if string(content) != "runtime: langgraph\ntier: 2\n" {
+		t.Errorf("config.yaml content: got %q", string(content))
+	}
+}
+
+func TestBuildBundleConfigFiles_SystemPromptAndConfigYaml(t *testing.T) {
+	b := &Bundle{
+		SystemPrompt: "Be concise.",
+		Prompts: map[string]string{
+			"config.yaml": "runtime: langgraph\n",
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	if n := len(files); n != 2 {
+		t.Fatalf("system-prompt + config.yaml: want 2 files, got %d", n)
+	}
+	if _, ok := files["system-prompt.md"]; !ok {
+		t.Error("missing system-prompt.md")
+	}
+	if _, ok := files["config.yaml"]; !ok {
+		t.Error("missing config.yaml")
+	}
+}
+
+func TestBuildBundleConfigFiles_Skills(t *testing.T) {
+	b := &Bundle{
+		Skills: []BundleSkill{
+			{
+				ID:   "web-search",
+				Files: map[string]string{"readme.md": "# Web Search\n"},
+			},
+			{
+				ID:   "code-interpreter",
+				Files: map[string]string{"readme.md": "# Code Interpreter\n"},
+			},
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	// 2 skills × 1 file each = 2 files
+	if n := len(files); n != 2 {
+		t.Fatalf("skills: want 2 files, got %d", n)
+	}
+	if _, ok := files["skills/web-search/readme.md"]; !ok {
+		t.Error("missing skills/web-search/readme.md")
+	}
+	if _, ok := files["skills/code-interpreter/readme.md"]; !ok {
+		t.Error("missing skills/code-interpreter/readme.md")
+	}
+}
+
+func TestBuildBundleConfigFiles_SkillSubPaths(t *testing.T) {
+	b := &Bundle{
+		Skills: []BundleSkill{
+			{
+				ID: "multi-file",
+				Files: map[string]string{
+					"readme.md":        "# Multi",
+					"instructions.txt": "Step 1, Step 2",
+				},
+			},
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	if n := len(files); n != 2 {
+		t.Fatalf("skill with sub-paths: want 2 files, got %d", n)
+	}
+	if _, ok := files["skills/multi-file/readme.md"]; !ok {
+		t.Error("missing skills/multi-file/readme.md")
+	}
+	if _, ok := files["skills/multi-file/instructions.txt"]; !ok {
+		t.Error("missing skills/multi-file/instructions.txt")
+	}
+}
+
+func TestBuildBundleConfigFiles_EmptySystemPrompt(t *testing.T) {
+	b := &Bundle{
+		SystemPrompt: "",
+		Prompts: map[string]string{
+			"config.yaml": "runtime: langgraph\n",
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	// Empty system-prompt should not produce a file
+	if n := len(files); n != 1 {
+		t.Errorf("empty system-prompt: want 1 file, got %d", n)
+	}
+}
+
+func TestBuildBundleConfigFiles_EmptyPrompts(t *testing.T) {
+	b := &Bundle{
+		Prompts: map[string]string{},
+	}
+	files := buildBundleConfigFiles(b)
+	if n := len(files); n != 0 {
+		t.Errorf("empty prompts map: want 0 files, got %d", n)
+	}
+}
+
+func TestBuildBundleConfigFiles_emptyBundle(t *testing.T) {
+	b := &Bundle{}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 0 {
+		t.Errorf("expected empty map for empty bundle, got %d entries", len(files))
+	}
+}
+
+func TestBuildBundleConfigFiles_systemPrompt(t *testing.T) {
+	b := &Bundle{SystemPrompt: "You are a helpful assistant."}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 1 {
+		t.Fatalf("expected 1 file, got %d", len(files))
+	}
+	if string(files["system-prompt.md"]) != "You are a helpful assistant." {
+		t.Errorf("unexpected system prompt content: %q", files["system-prompt.md"])
+	}
+}
+
+func TestBuildBundleConfigFiles_configYaml(t *testing.T) {
+	b := &Bundle{Prompts: map[string]string{
+		"config.yaml": "runtime: langgraph\nmodel: claude-sonnet-4-20250514\n",
+	}}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 1 {
+		t.Fatalf("expected 1 file, got %d", len(files))
+	}
+	if string(files["config.yaml"]) != "runtime: langgraph\nmodel: claude-sonnet-4-20250514\n" {
+		t.Errorf("unexpected config.yaml content: %q", files["config.yaml"])
+	}
+}
+
+func TestBuildBundleConfigFiles_systemPromptAndConfigYaml(t *testing.T) {
+	b := &Bundle{
+		SystemPrompt: "# System",
+		Prompts:     map[string]string{"config.yaml": "runtime: langgraph"},
+	}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 2 {
+		t.Fatalf("expected 2 files, got %d", len(files))
+	}
+	if _, ok := files["system-prompt.md"]; !ok {
+		t.Error("missing system-prompt.md")
+	}
+	if _, ok := files["config.yaml"]; !ok {
+		t.Error("missing config.yaml")
+	}
+}
+
+func TestBuildBundleConfigFiles_skills(t *testing.T) {
+	b := &Bundle{
+		Skills: []BundleSkill{
+			{
+				ID:          "web-search",
+				Name:        "Web Search",
+				Description: "Search the web",
+				Files:       map[string]string{"readme.md": "# Web Search"},
+			},
+			{
+				ID:          "code-runner",
+				Name:        "Code Runner",
+				Description: "Execute code",
+				Files:       map[string]string{"handler.py": "print('hello')"},
+			},
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 2 {
+		t.Fatalf("expected 2 skill files, got %d", len(files))
+	}
+
+	if content, ok := files["skills/web-search/readme.md"]; !ok {
+		t.Error("missing skills/web-search/readme.md")
+	} else if string(content) != "# Web Search" {
+		t.Errorf("unexpected readme.md: %q", content)
+	}
+
+	if _, ok := files["skills/code-runner/handler.py"]; !ok {
+		t.Error("missing skills/code-runner/handler.py")
+	}
+}
+
+func TestBuildBundleConfigFiles_skillsWithSubPaths(t *testing.T) {
+	b := &Bundle{
+		Skills: []BundleSkill{
+			{
+				ID:    "nested-skill",
+				Files: map[string]string{"src/main.py": "def main(): pass", "pyproject.toml": "[tool.foo]"},
+			},
+		},
+	}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 2 {
+		t.Fatalf("expected 2 files, got %d", len(files))
+	}
+	if _, ok := files["skills/nested-skill/src/main.py"]; !ok {
+		t.Error("missing skills/nested-skill/src/main.py")
+	}
+	if _, ok := files["skills/nested-skill/pyproject.toml"]; !ok {
+		t.Error("missing skills/nested-skill/pyproject.toml")
+	}
+}
+
+func TestBuildBundleConfigFiles_skipsEmptyPrompts(t *testing.T) {
+	b := &Bundle{Prompts: map[string]string{}}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 0 {
+		t.Errorf("expected 0 files for empty prompts map, got %d", len(files))
+	}
+}
+
+func TestBuildBundleConfigFiles_skipsMissingConfigYaml(t *testing.T) {
+	b := &Bundle{
+		SystemPrompt: "# My Prompt",
+		Prompts:      map[string]string{"other.yaml": "something: else"},
+	}
+	files := buildBundleConfigFiles(b)
+	if len(files) != 1 {
+		t.Fatalf("expected 1 file (system-prompt only), got %d", len(files))
+	}
+	if _, ok := files["config.yaml"]; ok {
+		t.Error("config.yaml should not be written when not in Prompts")
+	}
+}
+
+func TestNilIfEmpty_emptyString(t *testing.T) {
+	result := nilIfEmpty("")
+	if result != nil {
+		t.Errorf("expected nil for empty string, got %v", result)
+	}
+}
+
+func TestNilIfEmpty_nonEmptyString(t *testing.T) {
+	result := nilIfEmpty("hello")
+	if result == nil {
+		t.Fatal("expected non-nil result for non-empty string")
+	}
+	if result != "hello" {
+		t.Errorf("expected hello, got %q", result)
+	}
+}
+
+func TestNilIfEmpty_whitespaceString(t *testing.T) {
+	// Whitespace is not empty — nilIfEmpty only checks for zero-length
+	result := nilIfEmpty("   ")
+	if result == nil {
+		t.Error("expected non-nil for whitespace string")
+	} else if result != "   " {
+		t.Errorf("expected '   ', got %q", result)
+	}
+}
+
+func TestNilIfEmpty_EmptyString(t *testing.T) {
+	got := nilIfEmpty("")
+	if got != nil {
+		t.Errorf("nilIfEmpty(\"\"): want nil, got %v", got)
+	}
+}
+
+func TestNilIfEmpty_NonEmptyString(t *testing.T) {
+	got := nilIfEmpty("hello")
+	if got == nil {
+		t.Fatal("nilIfEmpty(\"hello\"): want \"hello\", got nil")
+	}
+	if s, ok := got.(string); !ok || s != "hello" {
+		t.Errorf("nilIfEmpty(\"hello\"): got %v (%T)", got, got)
+	}
+}
+
+func TestNilIfEmpty_Whitespace(t *testing.T) {
+	got := nilIfEmpty("   ")
+	if got == nil {
+		t.Fatal("nilIfEmpty(\"   \"): want \"   \", got nil (whitespace is not empty)")
+	}
+	if s, ok := got.(string); !ok || s != "   " {
+		t.Errorf("nilIfEmpty(\"   \"): got %v (%T)", got, got)
+	}
+}

workspace-server/internal/channels/manager.go

@@ -402,7 +402,7 @@ func (m *Manager) SendOutbound(ctx context.Context, channelID string, text strin
 		return err
 	}
 
-	adapter, ok := GetAdapter(ch.ChannelType)
+	adapter, ok := GetSendAdapter(ch.ChannelType)
 	if !ok {
 		return fmt.Errorf("no adapter for %s", ch.ChannelType)
 	}

workspace-server/internal/channels/registry.go

@@ -1,5 +1,7 @@
 package channels
 
+import "context"
+
 // Registry of all available channel adapters.
 // To add a new platform: implement ChannelAdapter, register here.
 var adapters = map[string]ChannelAdapter{
@@ -9,6 +11,27 @@ var adapters = map[string]ChannelAdapter{
 	"discord":  &DiscordAdapter{},
 }
 
+// SendAdapter is the subset of ChannelAdapter needed by SendOutbound.
+// Extracted so tests can inject a no-op/mock adapter without hitting real
+// platform APIs (Telegram Bot API, Slack API, etc.).
+type SendAdapter interface {
+	SendMessage(ctx context.Context, config map[string]interface{}, chatID string, text string) error
+}
+
+// getSendAdapter is the production implementation of GetSendAdapter —
+// returns the real registered adapter's SendMessage method.
+func getSendAdapter(channelType string) (SendAdapter, bool) {
+	a, ok := adapters[channelType]
+	if !ok {
+		return nil, false
+	}
+	return a, true
+}
+
+// GetSendAdapter returns the SendAdapter for a channel type.
+// Defaults to the real adapter; overridden by SetTestSendAdapter in tests.
+var GetSendAdapter = getSendAdapter
+
 // GetAdapter returns the adapter for a channel type.
 func GetAdapter(channelType string) (ChannelAdapter, bool) {
 	a, ok := adapters[channelType]