claude-ceo-assistant
007dabd29b
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
CI / Python Lint & Test (pull_request) Successful in 4s
CI / Detect changes (pull_request) Successful in 13s
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 14s
E2E Chat / detect-changes (pull_request) Successful in 12s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 12s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 3s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 3s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 6s
E2E API Smoke Test / detect-changes (pull_request) Successful in 18s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 18s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 6s
gate-check-v3 / gate-check (pull_request_target) Successful in 7s
Check migration collisions / Migration version collision check (pull_request) Successful in 26s
sop-checklist / review-refire (pull_request_target) Has been skipped
Harness Replays / detect-changes (pull_request) Successful in 13s
sop-checklist / all-items-acked (pull_request) acked: 7/7
sop-checklist / na-declarations (pull_request) N/A: (none)
sop-checklist / all-items-acked (pull_request_target) Successful in 5s
E2E Chat / E2E Chat (pull_request) Successful in 2s
sop-tier-check / tier-check (pull_request_target) Successful in 5s
qa-review / approved (pull_request_target) Failing after 12s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 1s
security-review / approved (pull_request_target) Failing after 12s
Harness Replays / Harness Replays (pull_request) Successful in 1s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 6s
CI / Canvas (Next.js) (pull_request) Successful in 8s
CI / Canvas Deploy Reminder (pull_request) Has been skipped
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m2s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 52s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1m1s
E2E Staging External Runtime / E2E Staging External Runtime (pull_request) Successful in 5m24s
CI / Platform (Go) (pull_request) Successful in 5m42s
CI / all-required (pull_request) Successful in 2s
sop-tier-check / tier-check (pull_request_review) Successful in 3s
Serve the latest post-mortem rescue bundle for a boot-failed/terminated
workspace so "why won't my agent boot" is answerable WITHOUT a live
instance. Powers the future canvas "Why did this fail?" panel.
Read-path decision (the key reviewer item):
Part 2 (feat/rfc742-rescue-capture) ships the bundle via internal/audit
(audit.Emit), which is stdout->Vector->Loki + a best-effort local JSONL
on the tenant container's EPHEMERAL rootfs — it does NOT persist to a
queryable DB table. Serving the read from Loki would require giving the
tenant process a Loki query client + obs read creds it deliberately must
not have. So this PR ADDS a minimal, per-tenant `rescue_bundles` table +
migration and persists the already-redacted bundle on capture, then
reads the latest row. No Loki-query creds added to the tenant.
What's added:
- migration 20260531000000_rescue_bundles (table + (workspace_id,
captured_at DESC, id DESC) index). Idempotent CREATE ... IF NOT
EXISTS; unique prefix, no collision.
- internal/rescue: Bundle/Section types + an injected PersistBundle
package var (leaf-safe, same pattern as RunRemote/Redact). Capture
now accumulates the redacted sections and persists ONE bundle row
after the per-section Loki ship — Loki behavior unchanged; persist is
best-effort + never disturbs the boot-failure path.
- internal/rescuestore: queryable store (Persist + GetLatest), org
scoped via `($2 = '' OR org_id = $2)`, per-section 64KiB clamp.
- handlers.RescueReadHandler: GET /workspaces/:id/rescue. 200 latest /
404 none / 503 store fault. Sections returned verbatim (already
redacted at capture; never re-shipped). Response section count
bounded.
- route registered on the WorkspaceAuth-guarded /workspaces/:id group,
next to /files/* and /exec. Org isolation = TenantGuard (routing) +
WorkspaceAuth (token bound to :id) + the store's MOLECULE_ORG_ID
filter, so a sibling org cannot read another org's bundle.
Tests (fake the store; sqlmock for the Postgres store):
returns latest, 404 when none, org-scoping (sibling org -> 404),
503 on store error, shape/redaction-preserved, section bound; capture
persists exactly once with redacted content, persist failure is
swallowed, no-store-wired still ships to Loki.
Dependency / merge order: branched from feat/rfc742-rescue-capture
(Part 2) because Capture's persist hook is extended here. Part 2 must
merge first (or be merged together) — this PR's rescue.go changes build
on Part 2's rescue package.
go build / go test / -tags=integration all green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>