fix-resume-waiting-operator — WorksGood live mirror

Metadata

Status	done
Assigned	`agent-909`
Agent identity	`fbb2d89fe44d6b761bcb4b67748ab3d417f29df70a65f8d9403308f5bb2bab78`
Created	2026-07-27T06:30:24.054894431+00:00
Started	2026-07-27T06:31:25.417653454+00:00
Completed	2026-07-27T07:10:08.064450513+00:00
Tokens	16837754 in / 34476 out

Description

Objective

Make the documented wg resume TASK command correctly resume an explicitly Waiting task, including fail-closed legacy message waits like prototype-clearer-tui, without graph surgery, false failure/incomplete retries, or message-triggered resurrection.

Required behavior

For Status::Waiting, wg resume TASK must atomically append a fenced lifecycle WaitSatisfied event with actor=Operator and an operator receipt, transition Waiting→Open, clear the active wait condition, disarm any message subscription, clear stale assignment, preserve the prior worktree/checkpoint and exact Pi session ID for continuation, and kick dispatch.
This is explicit operator authority only. Ordinary messages—including legacy_unbound messages—remain inert and cannot invoke this path.
Preserve existing paused-draft/subgraph wg resume behavior byte-for-byte. Waiting resume acts on only the named task unless a separately explicit supported scope is implemented and tested.
Repeating resume must be idempotent/actionable, never duplicate a generation, attempt, or process.
Update misleading help/quickstart text if needed.

Live recovery

After installing the fixed binary, do not mutate prototype-clearer-tui from this worker. Log that the operator may safely run wg resume prototype-clearer-tui --only; the chat agent will perform that explicit command after validation.

Validation

RED regression proves current wg resume rejects a Waiting legacy-unbound task.
Waiting→Open is represented by exactly one fenced WaitSatisfied lifecycle event with Operator authority.
Wait state/assignment are cleared while session ID, checkpoint, and worktree WIP remain preserved.
Legacy/unbound and ordinary messages remain inert before explicit operator resume.
Dispatcher launches exactly one resumed attempt; no retry_count increase and no failed/incomplete transition.
Paused draft and downstream resume tests remain unchanged.
cargo fmt --check, focused tests, clippy/build, install, and owned smoke pass.

## Objective
Make the documented `wg resume TASK` command correctly resume an explicitly Waiting task, including fail-closed legacy message waits like `prototype-clearer-tui`, without graph surgery, false failure/incomplete retries, or message-triggered resurrection.

## Required behavior
- For `Status::Waiting`, `wg resume TASK` must atomically append a fenced lifecycle `WaitSatisfied` event with actor=Operator and an operator receipt, transition Waiting→Open, clear the active wait condition, disarm any message subscription, clear stale assignment, preserve the prior worktree/checkpoint and exact Pi session ID for continuation, and kick dispatch.
- This is explicit operator authority only. Ordinary messages—including `legacy_unbound` messages—remain inert and cannot invoke this path.
- Preserve existing paused-draft/subgraph `wg resume` behavior byte-for-byte. Waiting resume acts on only the named task unless a separately explicit supported scope is implemented and tested.
- Repeating resume must be idempotent/actionable, never duplicate a generation, attempt, or process.
- Update misleading help/quickstart text if needed.

## Live recovery
After installing the fixed binary, do not mutate `prototype-clearer-tui` from this worker. Log that the operator may safely run `wg resume prototype-clearer-tui --only`; the chat agent will perform that explicit command after validation.

## Validation
- [ ] RED regression proves current `wg resume` rejects a Waiting legacy-unbound task.
- [ ] Waiting→Open is represented by exactly one fenced `WaitSatisfied` lifecycle event with Operator authority.
- [ ] Wait state/assignment are cleared while session ID, checkpoint, and worktree WIP remain preserved.
- [ ] Legacy/unbound and ordinary messages remain inert before explicit operator resume.
- [ ] Dispatcher launches exactly one resumed attempt; no retry_count increase and no failed/incomplete transition.
- [ ] Paused draft and downstream resume tests remain unchanged.
- [ ] `cargo fmt --check`, focused tests, clippy/build, install, and owned smoke pass.

Depends on

(none)

Required by

(none)

Log

2026-07-27T06:30:24.033716620+00:00 Task paused
2026-07-27T06:30:24.232839778+00:00 Task published
2026-07-27T06:30:25.653378558+00:00 Spawn preparation deferred before launch permit; rollback is complete and no circuit-breaker charge was recorded. Repair the reported checkout/configuration condition and retry: spawn transaction for agent-909 rolled back (task remains dispatchable; rollback diagnostics: complete): failed to establish isolated-worktree observer baseline: observer state already exists; use open_at for restart reconciliation
2026-07-27T06:31:25.603771155+00:00 Spawned by coordinator --executor pi --model openai-codex:gpt-5.6-sol --isolation required-worktree
2026-07-27T06:31:43.941669062+00:00 Starting implementation: inspect prior WIP and current resume/wait lifecycle paths, then add regression and fix.
2026-07-27T06:38:09.957202078+00:00 RED proven: commands::resume::tests::test_resume_waiting_legacy_unbound_task fails because current wg resume rejects Waiting task as 'not paused'. Implementing fenced operator path inline (no fanout: lifecycle + CLI tests are same-file sequential work).
2026-07-27T06:50:44.377432541+00:00 Implementation complete: atomic Operator WaitSatisfied receipt path, idempotent dispatch-race handling, legacy message inertness tests, exact prior Pi session reuse/fail-closed continuation, updated help, and owned installed-CLI smoke scenario. Focused resume + Pi tests and fmt pass.
2026-07-27T07:09:34.933523382+00:00 Validation PASS: RED regression captured old Waiting rejection; focused Waiting/legacy-unbound + bound-subscription tests pass; exact Pi-session reuse/fail-closed test passes; integration_resume and legacy-unbound coordinator coverage pass; cargo fmt --check, cargo clippy, cargo build, and serial full cargo test pass (pre-existing warnings only); cargo install --path . --locked completed; installed-CLI resume help/quickstart verified; owned resume_waiting_operator smoke passes with real daemon lifecycle, one resumed attempt, exact Pi session, and dirty worktree reuse. A prior parallel full-test invocation exposed unrelated shared-HOME/global-profile interference; serial full suite passed.
2026-07-27T07:09:34.990359969+00:00 Live recovery note (no mutation performed here): the operator/chat agent may safely run: wg resume prototype-clearer-tui --only
2026-07-27T07:10:04.148799528+00:00 Committed and pushed 51b33f0cc04bcf403bee5de08f9181dd6a744a38 on origin/wg/agent-909/fix-resume-waiting-operator; worktree clean and remote synchronized.
2026-07-27T07:10:08.064451033+00:00 Task marked as done