mercury-agent 0.23.2 → 0.24.0

package/docs/plans/p0-plan-0922/briefs/S6.md

@@ -0,0 +1,77 @@
+# Brief S6 — a killed scheduled run leaves a ledger row
+
+**Plan**: docs/plans/p0-plan-0922/plan.md   **Session**: S6   **Round**: 2   **Kind**: feature
+**Item slug**: ledger-misses-interrupted-runs
+
+> The only file this session reads from the plan. Self-contained: it never points at a sibling's brief. The `**Item slug**:` line is the key the execution commands grep for to detect that a brief is active — keep it exactly in that shape.
+
+## Launch
+
+/f-ops-plan work p0-plan-0922 S6
+
+## Run
+
+/f-feature-dev ledger-misses-interrupted-runs
+
+## Scope notes
+
+Follow the doc's nine-step sequence. Step 4 (the `src/core/boot-reconcile.ts` seam) must come before step 5. **S7 builds on this seam**: after you land it appends a chat-run reconciler to `BOOT_RECONCILERS`. So write the seam's contract (the export names, the call order in `main.ts` before `startScheduler`, `bootAt`) into your log's **Siblings must know**, as it is in the code and not the doc. Owner decisions: an interrupted occurrence keeps today's make-up behaviour (only the label changes), and no chat notice goes to the group (the creator's missed-run report is reworded). **Deploy needs: rebuild + schema migration.** `src/types.ts` is baked, so run `container/build.sh`, then restart in a gap. The migration rebuilds `task_runs` on the live DB: rehearse on a shadow snapshot first (release-gate gate 2), and paste the §4 evidence block.
+
+## Owned surfaces (predicted)
+
+- scripts/football-audit.ts
+- src/audit/ledger-queries.ts
+- src/config.ts
+- src/core/boot-reconcile.ts
+- src/core/operator-alerts.ts
+- src/core/routes/dashboard.ts
+- src/core/runtime.ts
+- src/core/task-scheduler.ts
+- src/main.ts
+- src/storage/db.ts
+- src/types.ts
+- tests/boot-reconcile.test.ts
+- tests/dashboard-task-last-run.test.ts
+- tests/ledger-queries.test.ts
+- tests/operator-alerts.test.ts
+- tests/scheduler-interrupted-shutdown.test.ts
+- tests/scheduler-missed-runs.test.ts
+- tests/task-run-ledger.test.ts
+- tests/task-runs-migration.test.ts
+
+## Do not touch
+
+- container/build.sh, src/agent/container-entry.ts, src/agent/container-env.ts, src/agent/container-runner.ts, src/cli/mrctl.ts, src/core/api.ts, src/core/auth.ts, src/core/caller-token.ts, src/core/routes/console.ts, src/core/routes/media.ts, src/core/routes/storage.ts, src/preflight/checks/roundtrip.ts, tests/api-chat-auth.test.ts, tests/api-container-credential.test.ts, tests/api.test.ts, tests/console-routes.test.ts, tests/container-runner-create-args.test.ts, tests/media-storage-route.test.ts, tests/mrctl-http.test.ts, tests/preflight-roundtrip.test.ts   (owned by S5 this plan)
+- src/adapters/whatsapp.ts, src/agent/container-error.ts, src/agent/container-runner.ts, src/bridges/whatsapp.ts, src/core/chat-run-journal.ts, src/core/handler.ts, src/core/instance-lock.ts, src/core/space-queue.ts, src/core/system-messages.ts, tests/chat-run-journal.test.ts, tests/container-error.test.ts, tests/handler.test.ts, tests/instance-lock.test.ts, tests/orphaned-run-recovery.e2e.test.ts, tests/shutdown.test.ts, tests/space-queue.test.ts, tests/system-messages.test.ts   (owned by S7 this plan)
+- src/core/debounce.ts, src/core/handler.ts, src/core/pending-ingress.ts, tests/debounce.test.ts, tests/handler.test.ts, tests/pending-ingress.test.ts, tests/shutdown.test.ts   (owned by S8 this plan)
+- Not ours (avishai tsabari / Avishai-Tsabari, last 14 d): `src/core/runtime.ts`, `src/main.ts`, `src/storage/db.ts`, `src/config.ts`, `src/agent/container-runner.ts`, `src/agent/container-entry.ts`, `src/preflight/checks/roundtrip.ts`: see plan.md "Not ours". The lane is cleared by Q1 (ours). Re-check with `/f-ops-plan next p0-plan-0922`.
+- Plan `sensitive-bugs-0922` fixes the sensitive-guard bugs in `src/core/runtime.ts` *after* this plan. Do not fix any of them here, even if you pass the code: `handleRawInput` → `checkSensitiveConnectionGuard`, the replay at ~1728, and the scheduler skip at ~1047.
+
+## Edges
+
+- start-after: none   (satisfied — `next` launched you)
+- merge-after: none
+
+## Overlap verdicts
+
+none
+
+> The command's preflight overlap check treats a hit whose branch AND every shared file match a verdict line as answered (c), prints one `PLAN-PASSED(overlap): …` line, and carries it into the final report. Any other hit prompts as usual. The merge-time check is never waived.
+
+## Decisions answered
+
+- Q1: ours (owner, 2026-09-22). `src/core` and `src/agent` are this plan's lane.
+
+## Ledger rules
+
+- `docs/ROADMAP.md`: never edit. Your command writes `docs/pending-updates/ledger-misses-interrupted-runs.md` as usual.
+- `docs/pending-verification.md`: append only under `## Plan p0-plan-0922 / S6`.
+- Any goal's `decisions.md`: never edit; put decisions in your log's "Siblings must know".
+
+## Environment
+
+- Read `CLAUDE.md` → Worktree Setup and Gotchas before opening the worktree. Nothing from them is copied here. `bun install` once in the fresh worktree; the check gate is `bun run check` (every link runs to completion — read every line).
+
+## Report
+
+Your command's final report is appended to `docs/plans/p0-plan-0922/sessions/S6.md` by its report step (path resolved through `.claude/scripts/resolve-docs-root.sh docs/plans/probe.md`). `work` verifies the append.

package/docs/plans/p0-plan-0922/briefs/S7.md

@@ -0,0 +1,79 @@
+# Brief S7 — a hard kill leaves no dangling ⏳ and no silent drop
+
+**Plan**: docs/plans/p0-plan-0922/plan.md   **Session**: S7   **Round**: 3   **Kind**: feature
+**Item slug**: orphaned-run-recovery
+
+> The only file this session reads from the plan. Self-contained: it never points at a sibling's brief. The `**Item slug**:` line is the key the execution commands grep for to detect that a brief is active — keep it exactly in that shape.
+
+## Launch
+
+/f-ops-plan work p0-plan-0922 S7
+
+## Run
+
+/f-feature-dev orphaned-run-recovery
+
+## Scope notes
+
+Follow the doc's sequence. S6 has landed by the time you launch, so steps 5–6 are no longer blocked. Read the seam from the code, not from the doc. `ChatRunHandle` lives in `src/core/chat-run-journal.ts` and deliberately **not** in `src/types.ts` (baked), and that keeps this deploy **restart-only**. Do not move it. The instance lock (step 1) must still let a shadow take its own lock beside the live bot, so rehearse that. **Deploy needs: restart.** Deploy in a gap: no `mercury.managed` container and no `bin/pi`. Then the live check the doc names: restart mid-answer, see the ⏳ cleared and one "please send it again". `src/core/handler.ts`, `tests/handler.test.ts` and `tests/shutdown.test.ts` are shared with S8, which starts after you.
+
+## Owned surfaces (predicted)
+
+- src/adapters/whatsapp.ts
+- src/agent/container-error.ts
+- src/agent/container-runner.ts
+- src/bridges/whatsapp.ts
+- src/core/boot-reconcile.ts
+- src/core/chat-run-journal.ts
+- src/core/handler.ts
+- src/core/instance-lock.ts
+- src/core/runtime.ts
+- src/core/space-queue.ts
+- src/core/system-messages.ts
+- src/main.ts
+- src/storage/db.ts
+- tests/chat-run-journal.test.ts
+- tests/container-error.test.ts
+- tests/handler.test.ts
+- tests/instance-lock.test.ts
+- tests/orphaned-run-recovery.e2e.test.ts
+- tests/shutdown.test.ts
+- tests/space-queue.test.ts
+- tests/system-messages.test.ts
+
+## Do not touch
+
+- container/build.sh, src/agent/container-entry.ts, src/agent/container-env.ts, src/cli/mrctl.ts, src/config.ts, src/core/api.ts, src/core/auth.ts, src/core/caller-token.ts, src/core/routes/console.ts, src/core/routes/media.ts, src/core/routes/storage.ts, src/preflight/checks/roundtrip.ts, tests/api-chat-auth.test.ts, tests/api-container-credential.test.ts, tests/api.test.ts, tests/console-routes.test.ts, tests/container-runner-create-args.test.ts, tests/media-storage-route.test.ts, tests/mrctl-http.test.ts, tests/preflight-roundtrip.test.ts   (owned by S5 this plan)
+- scripts/football-audit.ts, src/audit/ledger-queries.ts, src/config.ts, src/core/operator-alerts.ts, src/core/routes/dashboard.ts, src/core/task-scheduler.ts, src/types.ts, tests/boot-reconcile.test.ts, tests/dashboard-task-last-run.test.ts, tests/ledger-queries.test.ts, tests/operator-alerts.test.ts, tests/scheduler-interrupted-shutdown.test.ts, tests/scheduler-missed-runs.test.ts, tests/task-run-ledger.test.ts, tests/task-runs-migration.test.ts   (owned by S6 this plan)
+- src/core/debounce.ts, src/core/pending-ingress.ts, src/types.ts, tests/debounce.test.ts, tests/pending-ingress.test.ts   (owned by S8 this plan)
+- Not ours (avishai tsabari / Avishai-Tsabari, last 14 d): `src/core/runtime.ts`, `src/main.ts`, `src/storage/db.ts`, `src/config.ts`, `src/agent/container-runner.ts`, `src/agent/container-entry.ts`, `src/preflight/checks/roundtrip.ts`: see plan.md "Not ours". The lane is cleared by Q1 (ours). Re-check with `/f-ops-plan next p0-plan-0922`.
+- Plan `sensitive-bugs-0922` fixes the sensitive-guard bugs in `src/core/runtime.ts` *after* this plan. Do not fix any of them here, even if you pass the code: `handleRawInput` → `checkSensitiveConnectionGuard`, the replay at ~1728, and the scheduler skip at ~1047.
+
+## Edges
+
+- start-after: none   (satisfied — `next` launched you)
+- merge-after: land only after S6 is landed (it will be: S6 is also a start-after predecessor). Before step 5, re-read S6's `src/core/boot-reconcile.ts` as landed on `dev` and S6's log "Siblings must know". Check with `/f-ops-plan next p0-plan-0922` before the rebase step.
+
+## Overlap verdicts
+
+none
+
+> The command's preflight overlap check treats a hit whose branch AND every shared file match a verdict line as answered (c), prints one `PLAN-PASSED(overlap): …` line, and carries it into the final report. Any other hit prompts as usual. The merge-time check is never waived.
+
+## Decisions answered
+
+- Q1: ours (owner, 2026-09-22). `src/core` and `src/agent` are this plan's lane.
+
+## Ledger rules
+
+- `docs/ROADMAP.md`: never edit. Your command writes `docs/pending-updates/orphaned-run-recovery.md` as usual.
+- `docs/pending-verification.md`: append only under `## Plan p0-plan-0922 / S7`.
+- Any goal's `decisions.md`: never edit; put decisions in your log's "Siblings must know".
+
+## Environment
+
+- Read `CLAUDE.md` → Worktree Setup and Gotchas before opening the worktree. Nothing from them is copied here. `bun install` once in the fresh worktree; the check gate is `bun run check` (every link runs to completion — read every line).
+
+## Report
+
+Your command's final report is appended to `docs/plans/p0-plan-0922/sessions/S7.md` by its report step (path resolved through `.claude/scripts/resolve-docs-root.sh docs/plans/probe.md`). `work` verifies the append.

package/docs/plans/p0-plan-0922/briefs/S8.md

@@ -0,0 +1,69 @@
+# Brief S8 — a restart persists and replays pending debounced messages
+
+**Plan**: docs/plans/p0-plan-0922/plan.md   **Session**: S8   **Round**: 4   **Kind**: feature
+**Item slug**: wire-debounce-flush-to-shutdown-hook
+
+> The only file this session reads from the plan. Self-contained: it never points at a sibling's brief. The `**Item slug**:` line is the key the execution commands grep for to detect that a brief is active — keep it exactly in that shape.
+
+## Launch
+
+/f-ops-plan work p0-plan-0922 S8
+
+## Run
+
+/f-feature-dev wire-debounce-flush-to-shutdown-hook
+
+## Scope notes
+
+Follow the doc's sequence. The owner decided to **persist, not flush**: the queue is cancelled and containers killed before the shutdown hooks run, inside a 10 s budget. `Debouncer.close(sink)` (step 2) is the delicate part. Read `src/core/debounce.ts` whole first: the stale-heal path and the `.finally()` identity guard are what a naive close breaks. S7 has landed by now and changed `src/core/handler.ts`, the shutdown path in `runtime.ts` and `SpaceQueue` settle. Rebase onto it and re-read those before step 3. **Deploy needs: restart** (no baked file, no version change). Confirm that with the release-gate §1 classification at deploy.
+
+## Owned surfaces (predicted)
+
+- src/core/debounce.ts
+- src/core/handler.ts
+- src/core/pending-ingress.ts
+- src/core/runtime.ts
+- src/main.ts
+- src/storage/db.ts
+- src/types.ts
+- tests/debounce.test.ts
+- tests/handler.test.ts
+- tests/pending-ingress.test.ts
+- tests/shutdown.test.ts
+
+## Do not touch
+
+- container/build.sh, src/agent/container-entry.ts, src/agent/container-env.ts, src/agent/container-runner.ts, src/cli/mrctl.ts, src/config.ts, src/core/api.ts, src/core/auth.ts, src/core/caller-token.ts, src/core/routes/console.ts, src/core/routes/media.ts, src/core/routes/storage.ts, src/preflight/checks/roundtrip.ts, tests/api-chat-auth.test.ts, tests/api-container-credential.test.ts, tests/api.test.ts, tests/console-routes.test.ts, tests/container-runner-create-args.test.ts, tests/media-storage-route.test.ts, tests/mrctl-http.test.ts, tests/preflight-roundtrip.test.ts   (owned by S5 this plan)
+- scripts/football-audit.ts, src/audit/ledger-queries.ts, src/config.ts, src/core/boot-reconcile.ts, src/core/operator-alerts.ts, src/core/routes/dashboard.ts, src/core/task-scheduler.ts, tests/boot-reconcile.test.ts, tests/dashboard-task-last-run.test.ts, tests/ledger-queries.test.ts, tests/operator-alerts.test.ts, tests/scheduler-interrupted-shutdown.test.ts, tests/scheduler-missed-runs.test.ts, tests/task-run-ledger.test.ts, tests/task-runs-migration.test.ts   (owned by S6 this plan)
+- src/adapters/whatsapp.ts, src/agent/container-error.ts, src/agent/container-runner.ts, src/bridges/whatsapp.ts, src/core/boot-reconcile.ts, src/core/chat-run-journal.ts, src/core/instance-lock.ts, src/core/space-queue.ts, src/core/system-messages.ts, tests/chat-run-journal.test.ts, tests/container-error.test.ts, tests/instance-lock.test.ts, tests/orphaned-run-recovery.e2e.test.ts, tests/space-queue.test.ts, tests/system-messages.test.ts   (owned by S7 this plan)
+- Not ours (avishai tsabari / Avishai-Tsabari, last 14 d): `src/core/runtime.ts`, `src/main.ts`, `src/storage/db.ts`, `src/config.ts`, `src/agent/container-runner.ts`, `src/agent/container-entry.ts`, `src/preflight/checks/roundtrip.ts`: see plan.md "Not ours". The lane is cleared by Q1 (ours). Re-check with `/f-ops-plan next p0-plan-0922`.
+- Plan `sensitive-bugs-0922` fixes the sensitive-guard bugs in `src/core/runtime.ts` *after* this plan. Do not fix any of them here, even if you pass the code: `handleRawInput` → `checkSensitiveConnectionGuard`, the replay at ~1728, and the scheduler skip at ~1047.
+
+## Edges
+
+- start-after: none   (satisfied — `next` launched you)
+- merge-after: none
+
+## Overlap verdicts
+
+none
+
+> The command's preflight overlap check treats a hit whose branch AND every shared file match a verdict line as answered (c), prints one `PLAN-PASSED(overlap): …` line, and carries it into the final report. Any other hit prompts as usual. The merge-time check is never waived.
+
+## Decisions answered
+
+- Q1: ours (owner, 2026-09-22). `src/core` and `src/agent` are this plan's lane.
+
+## Ledger rules
+
+- `docs/ROADMAP.md`: never edit. Your command writes `docs/pending-updates/wire-debounce-flush-to-shutdown-hook.md` as usual.
+- `docs/pending-verification.md`: append only under `## Plan p0-plan-0922 / S8`.
+- Any goal's `decisions.md`: never edit; put decisions in your log's "Siblings must know".
+
+## Environment
+
+- Read `CLAUDE.md` → Worktree Setup and Gotchas before opening the worktree. Nothing from them is copied here. `bun install` once in the fresh worktree; the check gate is `bun run check` (every link runs to completion — read every line).
+
+## Report
+
+Your command's final report is appended to `docs/plans/p0-plan-0922/sessions/S8.md` by its report step (path resolved through `.claude/scripts/resolve-docs-root.sh docs/plans/probe.md`). `work` verifies the append.

package/docs/plans/p0-plan-0922/decisions.md

@@ -0,0 +1,9 @@
+# Decisions — plan p0-plan-0922
+
+> Answer in the `Answer` column. Everything blocked on a question stays blocked until this column is non-empty for it. `build` and `next` add rows; only you write answers. Delete nothing — an answered row is the audit trail.
+
+| ID | Blocks | Question | Options | Answer | Answered |
+|----|--------|----------|---------|--------|----------|
+| Q1 | S5, S6, S7, S8 | Lane: does avishai tsabari own `src/core` + `src/agent` (runtime.ts, main.ts, db.ts, config.ts, container-runner.ts; latest touch 2026-09-13)? | ours / theirs / split | ours: no worktree open, not mid-change | 2026-09-22 (owner) |
+
+> No owner-blocked questions at build. The Open Questions in each idea were design questions, and their planning sessions resolved them. Q1 was added at `--rebuild` by the lane measurement.

package/docs/plans/p0-plan-0922/plan.md

@@ -0,0 +1,215 @@
+# Plan: p0-plan-0922
+
+**Status**: Active
+**Created**: 2026-09-22   **Last updated**: 2026-09-22 (rebuilt after Round 0)
+**Landing mode**: pr
+**Max width**: 4
+**Inputs**: p0-plan-0922 require-caller-token-for-container-originated-requests sensitive-guard-one-read sensitive-opt-out-should-deny-not-block ledger-misses-interrupted-runs orphaned-run-recovery wire-debounce-flush-to-shutdown-hook (from docs/notes/2026-09-22-open-items-priority-scan.md P0 #2, #5, #6) · rebuilt 2026-09-22: `--rebuild`, adding the four implementation sessions S5–S8
+**Me**: saharw01, Saharw01@gmail.com (also 92163791+saharw01@users.noreply.github.com, avishailevi2@gmail.com)
+
+> Written by `/f-ops-plan` build (and `--rebuild`) and `archive` only. Nothing perishable lives here — no base sha, no other author's file list, no story counts; `next` measures those every time it runs. `plan-state.sh` parses the tables below by their header row, so keep the column order.
+
+> **Why this batch.** This is the idea half of the 2026-09-22 open-items scan's suggested round. Round 0 (S1–S4, frozen) planned the four ideas and landed in `83384ee`. The rebuild adds one implementation session per item (S5–S8). The bugs from the same scan are the archived sibling plan `p0-bugs-0922`. **Strictly serial.** S6, S7 and S8 all touch `src/core/runtime.ts`, `src/main.ts` and `src/storage/db.ts`. S5 shares `src/config.ts` with S6 and `src/agent/container-runner.ts` with S7. So every round has width 1, and that is measured, not chosen. **Sibling plan:** the five sensitive-guard bugs that bench-patterns-0915 S6 filed are plan `sensitive-bugs-0922`. The owner chose features first (2026-09-22), so that plan's sessions start after S8 here, by cross-plan edges it owns.
+
+## Items
+
+| ID | Kind | Slug | Doc | Command | Round | Needs |
+|----|------|------|-----|---------|-------|-------|
+| S1 | plan | require-caller-token-for-container-originated-requests | docs/ideas/require-caller-token-for-container-originated-requests.md | /f-feature-planning require-caller-token-for-container-originated-requests | 0 | — |
+| S2 | plan | ledger-misses-interrupted-runs | docs/ideas/ledger-misses-interrupted-runs.md | /f-feature-planning ledger-misses-interrupted-runs | 0 | — |
+| S3 | plan | orphaned-run-recovery | docs/ideas/orphaned-run-recovery.md | /f-feature-planning orphaned-run-recovery | 0 | — |
+| S4 | plan | wire-debounce-flush-to-shutdown-hook | docs/ideas/wire-debounce-flush-to-shutdown-hook.md | /f-feature-planning wire-debounce-flush-to-shutdown-hook | 0 | — |
+| S5 | feature | require-caller-token-for-container-originated-requests | docs/backlog/require-caller-token-for-container-originated-requests.md | /f-feature-dev require-caller-token-for-container-originated-requests | 1 | — |
+| S6 | feature | ledger-misses-interrupted-runs | docs/backlog/ledger-misses-interrupted-runs.md | /f-feature-dev ledger-misses-interrupted-runs | 2 | — |
+| S7 | feature | orphaned-run-recovery | docs/backlog/orphaned-run-recovery.md | /f-feature-dev orphaned-run-recovery | 3 | — |
+| S8 | feature | wire-debounce-flush-to-shutdown-hook | docs/backlog/wire-debounce-flush-to-shutdown-hook.md | /f-feature-dev wire-debounce-flush-to-shutdown-hook | 4 | — |
+
+> Kind ∈ `feature` | `bug` | `refactor` | `plan` (a Round 0 planning session) | `story` (from `--goal`; runs `/f-feature-dev`).
+> Needs = decision IDs from `decisions.md` that must be answered before this session may launch, or `—`.
+> IDs are never renumbered; `--drop` moves a row to Deferred, `--add` takes the next unused ID.
+
+## Edges
+
+| From | To | Kind | Basis |
+|------|----|------|-------|
+| S2 | S3 | merge-after | semantic: orphaned-run-recovery's startup sweep builds on the ledger's insert-at-start / reconcile-on-boot row (user) |
+| S5 | S6 | start-after | file: src/config.ts (both predicted) |
+| S5 | S7 | start-after | file: src/agent/container-runner.ts (both predicted) |
+| S6 | S7 | start-after | file: src/core/runtime.ts (both predicted; also src/main.ts, src/storage/db.ts, src/core/boot-reconcile.ts) |
+| S6 | S8 | start-after | file: src/core/runtime.ts (both predicted; also src/main.ts, src/storage/db.ts, src/types.ts) |
+| S7 | S8 | start-after | file: src/core/runtime.ts (both predicted; also src/main.ts, src/storage/db.ts, src/core/handler.ts, tests/handler.test.ts, tests/shutdown.test.ts) |
+| S6 | S7 | merge-after | semantic: orphaned-run-recovery's reconciler (its steps 5–6) appends to the `BOOT_RECONCILERS` seam that ledger-misses-interrupted-runs creates in `src/core/boot-reconcile.ts` (user) |
+
+> Kind ∈ `start-after` (To may not launch until From is landed) | `merge-after` (To may launch now, may not land until From is landed).
+> Basis starts with `file:` (inferred from the collision map) or `semantic:` (asked) — never mixed in one row.
+> From may name a session of ANOTHER active plan as `{plan-slug}/S{n}` — always `start-after`, always `file:`; resolved at `next` time by deriving that plan's state.
+
+## Rounds
+
+### Round 0 — unblock (landed; committed in 83384ee)
+
+| Session | Action | Clears |
+|---------|--------|--------|
+| S1 | /f-feature-planning require-caller-token-for-container-originated-requests | S5 |
+| S2 | /f-feature-planning ledger-misses-interrupted-runs | S6 |
+| S3 | /f-feature-planning orphaned-run-recovery | S7 |
+| S4 | /f-feature-planning wire-debounce-flush-to-shutdown-hook | S8 |
+
+### Round 1 — width 1
+
+| Session | Action | Clears |
+|---------|--------|--------|
+| S5 | /f-feature-dev require-caller-token-for-container-originated-requests | S6, S7 (src/config.ts, container-runner.ts) |
+
+### Round 2 — width 1
+
+| Session | Action | Clears |
+|---------|--------|--------|
+| S6 | /f-feature-dev ledger-misses-interrupted-runs | S7, S8 |
+
+### Round 3 — width 1
+
+| Session | Action | Clears |
+|---------|--------|--------|
+| S7 | /f-feature-dev orphaned-run-recovery | S8 |
+
+### Round 4 — width 1
+
+| Session | Action | Clears |
+|---------|--------|--------|
+| S8 | /f-feature-dev wire-debounce-flush-to-shutdown-hook | sensitive-bugs-0922 (its cross-plan edges) |
+
+## Collision map
+
+> Whole files, predicted; `docs/` and infrastructure paths are excluded exactly as `overlap-check.sh` excludes them (the predictor's `docs/…` hits, bare basenames that duplicate a full path, and host paths such as `~/sync-mercury.sh`, `.env` and `bin/pi` are ignored: 38 such paths, 110 predicted → 72 kept). Line ranges are folded into their file.
+
+| File | Sessions |
+|------|----------|
+| src/core/runtime.ts | S6, S7, S8 |
+| src/main.ts | S6, S7, S8 |
+| src/storage/db.ts | S6, S7, S8 |
+| src/config.ts | S5, S6 |
+| src/agent/container-runner.ts | S5, S7 |
+| src/core/boot-reconcile.ts (new) | S6, S7 |
+| src/types.ts | S6, S8 |
+| src/core/handler.ts | S7, S8 |
+| tests/handler.test.ts | S7, S8 |
+| tests/shutdown.test.ts | S7, S8 |
+
+## Predicted surfaces
+
+### S1
+
+### S2
+
+### S3
+
+### S4
+
+### S5
+
+- container/build.sh
+- src/agent/container-entry.ts
+- src/agent/container-env.ts
+- src/agent/container-runner.ts
+- src/cli/mrctl.ts
+- src/config.ts
+- src/core/api.ts
+- src/core/auth.ts
+- src/core/caller-token.ts
+- src/core/routes/console.ts
+- src/core/routes/media.ts
+- src/core/routes/storage.ts
+- src/preflight/checks/roundtrip.ts
+- tests/api-chat-auth.test.ts
+- tests/api-container-credential.test.ts
+- tests/api.test.ts
+- tests/console-routes.test.ts
+- tests/container-runner-create-args.test.ts
+- tests/media-storage-route.test.ts
+- tests/mrctl-http.test.ts
+- tests/preflight-roundtrip.test.ts
+
+### S6
+
+- scripts/football-audit.ts
+- src/audit/ledger-queries.ts
+- src/config.ts
+- src/core/boot-reconcile.ts
+- src/core/operator-alerts.ts
+- src/core/routes/dashboard.ts
+- src/core/runtime.ts
+- src/core/task-scheduler.ts
+- src/main.ts
+- src/storage/db.ts
+- src/types.ts
+- tests/boot-reconcile.test.ts
+- tests/dashboard-task-last-run.test.ts
+- tests/ledger-queries.test.ts
+- tests/operator-alerts.test.ts
+- tests/scheduler-interrupted-shutdown.test.ts
+- tests/scheduler-missed-runs.test.ts
+- tests/task-run-ledger.test.ts
+- tests/task-runs-migration.test.ts
+
+### S7
+
+- src/adapters/whatsapp.ts
+- src/agent/container-error.ts
+- src/agent/container-runner.ts
+- src/bridges/whatsapp.ts
+- src/core/boot-reconcile.ts
+- src/core/chat-run-journal.ts
+- src/core/handler.ts
+- src/core/instance-lock.ts
+- src/core/runtime.ts
+- src/core/space-queue.ts
+- src/core/system-messages.ts
+- src/main.ts
+- src/storage/db.ts
+- tests/chat-run-journal.test.ts
+- tests/container-error.test.ts
+- tests/handler.test.ts
+- tests/instance-lock.test.ts
+- tests/orphaned-run-recovery.e2e.test.ts
+- tests/shutdown.test.ts
+- tests/space-queue.test.ts
+- tests/system-messages.test.ts
+
+### S8
+
+- src/core/debounce.ts
+- src/core/handler.ts
+- src/core/pending-ingress.ts
+- src/core/runtime.ts
+- src/main.ts
+- src/storage/db.ts
+- src/types.ts
+- tests/debounce.test.ts
+- tests/handler.test.ts
+- tests/pending-ingress.test.ts
+- tests/shutdown.test.ts
+
+## Adjudicated overlaps
+
+> Pre-answers for the execution commands' preflight overlap prompt. Scoped to exact branch + file pairs on branches that belong to NO active plan (a hand-opened worktree, another contributor). A branch of a sibling plan's session is never listed here — it becomes a cross-plan edge instead.
+
+| Session | Branch (not in this plan) | Files | Verdict | Decided |
+|---------|---------------------------|-------|---------|---------|
+
+> Rebuild measurement 2026-09-22: no worktree other than the primary checkout, and no `worktree-*` branch. Nothing to adjudicate.
+
+## Not ours
+
+> Last measurement (`git log --since=14.days`, authors not in `Me`). Re-measured on every `next`; this table is the snapshot the briefs were written from.
+
+| File / area | Author | Basis |
+|-------------|--------|-------|
+| src/core/runtime.ts, src/main.ts, src/storage/db.ts, src/config.ts | avishai tsabari / Avishai-Tsabari | 14-day log (latest `eaf58ba` 2026-09-13, /spaces link); S6, S7, S8 (and S5 for config.ts). Lane: Q1 = ours |
+| src/agent/container-runner.ts, src/agent/container-entry.ts, src/preflight/checks/roundtrip.ts, tests/container-runner-create-args.test.ts, tests/preflight-roundtrip.test.ts | avishai tsabari / Avishai-Tsabari | 14-day log (2026-09-11 container fixes); S5, S7. Lane: Q1 = ours |
+
+## Deferred
+
+| Item | Reason | Home |
+|------|--------|------|
+| sensitive-opt-out-should-deny-not-block | moved to plan sensitive-bugs-0922 (rebuild 2026-09-22): the guard read landed; the opt-out now waits for B1 `sensitive-gate-is-checked-before-the-space-queue`, which that plan holds | docs/ideas/sensitive-opt-out-should-deny-not-block.md |

package/docs/plans/p0-plan-0922/sessions/S1.md

@@ -0,0 +1,22 @@
+## Launched — 2026-09-22T19:30Z
+Branch: —
+
+## Report — 2026-09-22T21:05Z
+Outcome: delivered
+Delivered: local (uncommitted — shared checkout; committed with Round 0)
+Blocked-on: —
+Deploy needs: none
+Files changed:
+- docs/backlog/require-caller-token-for-container-originated-requests.md
+- docs/ideas/require-caller-token-for-container-originated-requests.md (deleted — promoted)
+- docs/ROADMAP.md (Next row 14)
+Siblings must know:
+- none
+--- final report (verbatim) ---
+~~~
+Not captured. The owner ran /f-feature-planning require-caller-token-for-container-originated-requests directly, not through
+/f-ops-plan work, so this block was backfilled at cleanup (2026-09-22). Launch
+time is approximate. Outcome is delivered because
+docs/backlog/require-caller-token-for-container-originated-requests.md predicts surfaces and the owner approved its
+architecture.
+~~~

package/docs/plans/p0-plan-0922/sessions/S2.md

@@ -0,0 +1,22 @@
+## Launched — 2026-09-22T19:30Z
+Branch: —
+
+## Report — 2026-09-22T21:05Z
+Outcome: delivered
+Delivered: local (uncommitted — shared checkout; committed with Round 0)
+Blocked-on: —
+Deploy needs: none
+Files changed:
+- docs/backlog/ledger-misses-interrupted-runs.md
+- docs/ideas/ledger-misses-interrupted-runs.md (deleted — promoted)
+- docs/ROADMAP.md (Next row 15)
+Siblings must know:
+- none
+--- final report (verbatim) ---
+~~~
+Not captured. The owner ran /f-feature-planning ledger-misses-interrupted-runs directly, not through
+/f-ops-plan work, so this block was backfilled at cleanup (2026-09-22). Launch
+time is approximate. Outcome is delivered because
+docs/backlog/ledger-misses-interrupted-runs.md predicts surfaces and the owner approved its
+architecture.
+~~~

package/docs/plans/p0-plan-0922/sessions/S3.md

@@ -0,0 +1,22 @@
+## Launched — 2026-09-22T19:30Z
+Branch: —
+
+## Report — 2026-09-22T21:05Z
+Outcome: delivered
+Delivered: local (uncommitted — shared checkout; committed with Round 0)
+Blocked-on: —
+Deploy needs: none
+Files changed:
+- docs/backlog/orphaned-run-recovery.md
+- docs/ideas/orphaned-run-recovery.md (deleted — promoted)
+- docs/ROADMAP.md (Next row 16)
+Siblings must know:
+- none
+--- final report (verbatim) ---
+~~~
+Not captured. The owner ran /f-feature-planning orphaned-run-recovery directly, not through
+/f-ops-plan work, so this block was backfilled at cleanup (2026-09-22). Launch
+time is approximate. Outcome is delivered because
+docs/backlog/orphaned-run-recovery.md predicts surfaces and the owner approved its
+architecture.
+~~~

package/docs/plans/p0-plan-0922/sessions/S4.md

@@ -0,0 +1,22 @@
+## Launched — 2026-09-22T19:30Z
+Branch: —
+
+## Report — 2026-09-22T21:05Z
+Outcome: delivered
+Delivered: local (uncommitted — shared checkout; committed with Round 0)
+Blocked-on: —
+Deploy needs: none
+Files changed:
+- docs/backlog/wire-debounce-flush-to-shutdown-hook.md
+- docs/ideas/wire-debounce-flush-to-shutdown-hook.md (deleted — promoted)
+- docs/ROADMAP.md (Next row 17)
+Siblings must know:
+- none
+--- final report (verbatim) ---
+~~~
+Not captured. The owner ran /f-feature-planning wire-debounce-flush-to-shutdown-hook directly, not through
+/f-ops-plan work, so this block was backfilled at cleanup (2026-09-22). Launch
+time is approximate. Outcome is delivered because
+docs/backlog/wire-debounce-flush-to-shutdown-hook.md predicts surfaces and the owner approved its
+architecture.
+~~~

package/docs/plans/p0-plan-0922/sessions/S5.md

@@ -0,0 +1,44 @@
+## Launched — 2026-09-22T22:04Z
+Branch: worktree-feature-require-caller-token-for-container-originated-requests
+
+## Report — 2026-09-23T10:44Z
+Outcome: delivered
+Delivered: —
+Blocked-on: —
+Deploy needs: rebuild — container/build.sh, restart in a proven gap, wait for "Built derived agent image" (mrctl.ts + mrctl-http.ts are baked)
+Files changed:
+- docs/archive/permissions-and-config/2026-09-23-require-caller-token-for-container-originated-requests.md
+- docs/archive/applicative-profiles/2026-07-01-caller-bound-capability-token.md
+- docs/container-lifecycle.md
+- docs/ideas/rotate-api-secret-after-caller-token-cutover.md
+- docs/pending-updates/require-caller-token-for-container-originated-requests.md
+- src/agent/container-env.ts
+- src/agent/container-runner.ts
+- src/cli/mrctl-http.ts
+- src/cli/mrctl.ts
+- src/core/api.ts
+- src/core/routes/media.ts
+- src/preflight/checks/roundtrip.ts
+- tests/api-container-credential.test.ts
+- tests/container-runner-create-args.test.ts
+- tests/destructive-command-guards.test.ts
+- tests/media-storage-route.test.ts
+- tests/mrctl-http.test.ts
+- tests/preflight-roundtrip.test.ts
+Siblings must know:
+- `buildAgentCreateArgs` now routes every env pair through `pushEnv`, which drops any `-e` whose `KEY=value` contains a host secret (`hostSecretValues(config)` in `src/agent/container-env.ts`) and reports the key via `onSecretStripped`. S6/S7 editing `container-runner.ts`: add env through `envPairs`/`extraEnv`, never a bare `args.push("-e", …)`.
+- `reply()` no longer passes `API_SECRET`; `CALLER_TOKEN` is a container's only credential. Anything in-container that calls `/api` must send `x-mercury-token` (no bearer needed); `/api/console/*`, `/dashboard/*`, `/chat`, `/pre-build-ext-image` are unreachable from a container.
+- `/api` identity is now trustworthy: a present token decides it, headers only behind the operator secret. sensitive-bugs-0922 S1/S5 (gated on this session) can narrow on `auth.callerId` without re-checking for header forgery.
+- `src/config.ts` was NOT touched (no new key).
+--- final report (verbatim) ---
+~~~
+S5 require-caller-token-for-container-originated-requests — squashed to LOCAL dev as 7aecf70, NOT pushed (owner approval required per CLAUDE.local.md; landing derives once the archive is on origin/dev).
+- /api: two principals. Present x-mercury-token (empty counts) → verified or 401, never a fall-through; identity from the token. No token → Bearer API_SECRET + identity headers, logged "API request on the operator credential". 503 when no secret configured, either principal.
+- reply() no longer passes API_SECRET; buildAgentCreateArgs strips any -e pair whose value contains apiSecret/chatApiKey/callerTokenKey/shadowApiSecret (envPairs and extension env), logging the key at error.
+- mrctl: refuses with no credential (exit 1, "no credential — CALLER_TOKEN is not set …"); `mrctl disk` → GET /api/media/storage behind media.purge (admin-only — behaviour change for members).
+- Preflight: token mode presents the caller token alone; headers mode is the operator path and its pass title says the token leg was not exercised.
+- Docs: container-lifecycle two-principal section; archived caller-bound-token design carries a dated correction; residual idea docs/ideas/rotate-api-secret-after-caller-token-cutover.md (not before one week after live proof).
+- Gate: ~/check-mercury.sh check → 12/12 links, 5829 pass / 9 skip / 0 fail. Every new guard negative-tested (deleted → red). Code review: no bugs, 3 small fixes applied. QA: 24/30 pass, 6 deferred MANUAL-VERIFY (live only), 0 fail.
+- Deviations: branched from local dev (origin/dev lacked 2 local docs commits and the promotion — nothing pushed); tests/destructive-command-guards.test.ts needed CALLER_TOKEN in its stub mrctl env; mrctl-http.ts (baked) gained credentialHeaders.
+- PLAN-PASSED(overlap): none — preflight overlap check was clean (exit 0).
+~~~