fix-chat-create-obsolete-daemon — WorksGood live mirror

Metadata

Status	done
Agent identity	`fbb2d89fe44d6b761bcb4b67748ab3d417f29df70a65f8d9403308f5bb2bab78`
Created	2026-08-01T06:58:21.490564352+00:00
Started	2026-08-01T06:58:29.760098703+00:00
Completed	2026-08-03T09:42:51.327516945+00:00
Tokens	24122718 in / 47174 out

Description

Fix the attended TUI chat-create failure observed in /home/bot/jobs: a current TUI was correctly pinned to /home/bot/jobs/.wg, but that graph still had a 38-day-old live daemon process running an obsolete binary without the dedicated chat.sock. The compatibility fallback to the general IPC lane timed out; request-id reconciliation found no graph/runtime receipt, and the TUI showed chat creation did not produce a reconcilable graph commit; no chat/runtime residue was found. Restarting only that project daemon on the current binary immediately fixed creation and produced exactly one .chat-1 pinned to pi:zai:glm-5.2.

Make this failure self-explanatory and recoverable without cross-graph mutation. A current client/TUI must authenticate the explicit graph/service identity and detect an obsolete/unresponsive daemon before or after a lost create response. Prefer a safe idempotent same-request recovery: when policy permits and no ownership ambiguity exists, restart/reconnect the exact project daemon and retry the same request ID; otherwise show a precise wg --dir <exact> service restart action. Never silently connect to an inherited WG_DIR, never create in another graph, never allocate a duplicate chat, never kill an unverified/reused PID, and never claim success without the durable graph receipt. Preserve route-less attended Pi selection and existing chat/session history.

Validation

Candidate-binary test with current TUI/client + explicit graph + old/compat daemon lacking chat.sock; no generic unreconcilable error.
Lost response after commit returns the exact existing chat by request ID.
Lost response before commit either safely restarts/retries the same request or emits precise project-scoped recovery; zero residue/duplicates.
Inherited sibling WG_DIR cannot redirect graph, socket, receipt lookup, or restart target.
PID reuse/unverified legacy state cannot authorize killing an unrelated process.
Existing chats and route generation remain unchanged.
cargo fmt --check, focused chat IPC/TUI/service-identity tests, cargo check, cargo clippy.

Fix the attended TUI chat-create failure observed in `/home/bot/jobs`: a current TUI was correctly pinned to `/home/bot/jobs/.wg`, but that graph still had a 38-day-old live daemon process running an obsolete binary without the dedicated `chat.sock`. The compatibility fallback to the general IPC lane timed out; request-id reconciliation found no graph/runtime receipt, and the TUI showed `chat creation did not produce a reconcilable graph commit; no chat/runtime residue was found`. Restarting only that project daemon on the current binary immediately fixed creation and produced exactly one `.chat-1` pinned to `pi:zai:glm-5.2`.

Make this failure self-explanatory and recoverable without cross-graph mutation. A current client/TUI must authenticate the explicit graph/service identity and detect an obsolete/unresponsive daemon before or after a lost create response. Prefer a safe idempotent same-request recovery: when policy permits and no ownership ambiguity exists, restart/reconnect the exact project daemon and retry the same request ID; otherwise show a precise `wg --dir <exact> service restart` action. Never silently connect to an inherited WG_DIR, never create in another graph, never allocate a duplicate chat, never kill an unverified/reused PID, and never claim success without the durable graph receipt. Preserve route-less attended Pi selection and existing chat/session history.

## Validation
- Candidate-binary test with current TUI/client + explicit graph + old/compat daemon lacking `chat.sock`; no generic unreconcilable error.
- Lost response after commit returns the exact existing chat by request ID.
- Lost response before commit either safely restarts/retries the same request or emits precise project-scoped recovery; zero residue/duplicates.
- Inherited sibling `WG_DIR` cannot redirect graph, socket, receipt lookup, or restart target.
- PID reuse/unverified legacy state cannot authorize killing an unrelated process.
- Existing chats and route generation remain unchanged.
- cargo fmt --check, focused chat IPC/TUI/service-identity tests, cargo check, cargo clippy.

Depends on

done .assign-fix-chat-create-obsolete-daemon

Required by

(none)

✉ Messages 1 message (1 unread)

#1fix-chat-create-obsolete-daemon2026-08-01T07:52:19.408709269+00:00read

Validation complete: owned candidate-binary TUI obsolete-daemon smoke PASS; service unit suite 69/69 PASS; service_identity tests 2/2 PASS; cargo fmt/check/clippy PASS (repository-existing warnings only); cargo install --path . --locked PASS. Existing timeout-reconcile smoke reached all chat/IPC checks but its unrelated Ctrl+O footer-width assertion was clipped at 180 columns on rerun.

Log

2026-08-01T06:58:21.429280593+00:00 Task paused
2026-08-01T06:58:22.092795819+00:00 Task published
2026-08-01T06:58:29.359414794+00:00 Lightweight assignment: agent=Pi Lifecycle Hotfix Programmer (fbb2d89f), exec_mode=full, context_scope=graph, reason=Complex lifecycle recovery spans TUI, IPC, service identity, PID safety, and durable receipt reconciliation, so the lifecycle hotfix programmer with event-exclusion safety is the closest fit.
2026-08-01T06:58:30.001872537+00:00 Spawned by coordinator --executor pi --model openai-codex:gpt-5.6-sol --isolation required-worktree
2026-08-01T06:58:45.240292816+00:00 Starting implementation; inspecting prior WIP and chat/service IPC identity paths.
2026-08-01T07:35:15.488823941+00:00 Implemented exact-graph service identity preflight, authenticated obsolete-build controlled restart, durable receipt verification/same-ID retry, unreceipted-residue refusal, PID-birth signal fences, dedicated status identity lane, and route/history preservation across restart.
2026-08-01T07:53:23.521075616+00:00 Bounded evaluator infrastructure state error[WG-EVAL-INSUFFICIENT-EVIDENCE]: bounded evidence insufficient [candidate-manifest:candidate-manifest:truncated,candidate-source:candidate-source:truncated,declared-artifact:declared-artifact-000:truncated,declared-artifact:declared-artifact-001:truncated,declared-artifact:declared-artifact-002:truncated,declared-artifact:declared-artifact-003:truncated,declared-artifact:declared-artifact-004:truncated,declared-artifact:declared-artifact-005:truncated] without semantic rejection or cross-executor fallback; source remains in-progress
2026-08-01T07:54:01.239754383+00:00 Consumed deep-readonly FLIP report deep-report-a286c0961944d38733c4550fd5fc03bf4d4151f673e7b92e41ad0c1022aff2d0; observations=18 findings=3 route=b3:07cbfd6e71528de4a03ba3051159b1a811230a2a0a97c83e9fcb743edd0c2466 usage=63755in/1394out
2026-08-01T07:54:06.568316623+00:00 Bounded evaluator infrastructure state error[WG-EVAL-EVIDENCE-UNAVAILABLE]: automatic immutable evidence assembly failed without semantic rejection or cross-executor fallback; source remains in-progress
2026-08-01T07:54:38.903242509+00:00 Bounded evaluator infrastructure state error[WG-EVAL-EVIDENCE-UNAVAILABLE]: automatic immutable evidence assembly failed without semantic rejection or cross-executor fallback; source remains in-progress