• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

Best possible solution:

Land the focused recovery path after maintainer review accepts the side-effect-aware replay semantics and keeps the deterministic Crabbox plus live Codex proof attached to the PR verification record.

Do we have a high-confidence way to reproduce the issue?

Yes. The linked issue has live Discord/Telegram logs for item/completed followed by no turn/completed, and the PR supplies deterministic gateway proof of the same boundary; I did not execute tests in this read-only review.

Is this the best way to solve the issue?

Yes as a bounded first fix. The PR avoids automatic replay after possible side effects, surfaces a specific timeout outcome, and leaves broader automatic resume policy to maintainer judgment.

Label changes:

• add P1: The PR targets an urgent Codex app-server stall affecting user-facing agent/channel workflows where turns can abort after productive work.
• add merge-risk: 🚨 session-state: The diff deliberately changes replay safety and liveness metadata for timed-out Codex turns, which affects how session recovery treats prior tool activity.
• add proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes after-fix AWS Crabbox deterministic gateway proof and live Codex provider sanity output with concrete run IDs and observed results.
• add rating: 🐚 platinum hermit: Current PR rating is 🐚 platinum hermit because proof is 🦞 diamond lobster, patch quality is 🐚 platinum hermit, and Strong proof and focused tests support the patch, while the changed session-state semantics still need maintainer acceptance before merge.
• add status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): The PR body includes after-fix AWS Crabbox deterministic gateway proof and live Codex provider sanity output with concrete run IDs and observed results.

Label justifications:

• P1: The PR targets an urgent Codex app-server stall affecting user-facing agent/channel workflows where turns can abort after productive work.
• merge-risk: 🚨 session-state: The diff deliberately changes replay safety and liveness metadata for timed-out Codex turns, which affects how session recovery treats prior tool activity.
• rating: 🐚 platinum hermit: Current PR rating is 🐚 platinum hermit because proof is 🦞 diamond lobster, patch quality is 🐚 platinum hermit, and Strong proof and focused tests support the patch, while the changed session-state semantics still need maintainer acceptance before merge.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): The PR body includes after-fix AWS Crabbox deterministic gateway proof and live Codex provider sanity output with concrete run IDs and observed results.
• proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes after-fix AWS Crabbox deterministic gateway proof and live Codex provider sanity output with concrete run IDs and observed results.

What I checked:

• Protected PR state: The provided GitHub context shows this PR is open, mergeable, contributor-authored, and carries the protected maintainer label, so this workflow must keep it open for explicit maintainer handling.
• Current main timeout behavior: Current main aborts the Codex app-server run on completion-idle timeout after waiting for turn/completed, which matches the linked issue's reported failure boundary. (extensions/codex/src/app-server/run-attempt.ts:1553, 0ab1449215f5)
• PR timeout outcome: The PR head computes a promptTimeoutOutcome for completion-idle timeouts and marks side-effecting cases as replayInvalid: true with livenessState: "abandoned". (extensions/codex/src/app-server/run-attempt.ts:2687, ecb61d596278)
• Side-effect tracking: The PR head broadens replay metadata from messaging-tool-only evidence to include cron adds, native mutating/MCP tool items, and executed dynamic tools while preserving blocked and pre-execution failures as retry-safe. (extensions/codex/src/app-server/event-projector.ts:328, ecb61d596278)
• Focused regression coverage: The PR adds tests for completed native command timeouts, executed dynamic-tool timeouts, active mutating items, assistant-only timeout messaging, and runner propagation of harness-provided timeout metadata. (extensions/codex/src/app-server/run-attempt.test.ts:2855, ecb61d596278)
• Real behavior proof in PR body: The PR body reports AWS Crabbox deterministic E2E proof where the first turn timed out with replayInvalid=true and livenessState=abandoned, the second turn returned SECOND_RUN_OK, and a live Codex provider Docker harness passed. (ecb61d596278)

Likely related people:

• funmerlin: Authored the merged quiescent app-server turn fix that added the existing fail-fast behavior for last-item completion without turn/completed. (role: adjacent fix author; confidence: high; commits: 127156a88a29; files: extensions/codex/src/app-server/run-attempt.ts, extensions/codex/src/app-server/run-attempt.test.ts)
• joshavant: Authored the earlier merged Codex app-server liveness fix for account/rate-limit notifications and duplicate timeout suppression, and this PR builds on that same surface. (role: prior area contributor and current PR author; confidence: high; commits: 5fdef4c39e75, 437f322aafed, ecb61d596278; files: extensions/codex/src/app-server/run-attempt.ts, src/agents/pi-embedded-runner/run.ts)
• steipete: Recent GitHub file history shows repeated Codex app-server and embedded-runner changes, including current-day edits on run-attempt.ts and broader runtime/refactor work near this path. (role: recent area contributor; confidence: medium; commits: 1d5b5db4d221, 02182d5a3031, 4ff28a77356f; files: extensions/codex/src/app-server/run-attempt.ts, src/agents/pi-embedded-runner/run.ts)

Codex review notes: model gpt-5.5, reasoning high; reviewed against 0ab1449215f5.

• Eggs appear after the PR passes real-behavior proof. It is here for vibes, not verdicts: it does not change labels, ratings, merge decisions, or automation.
• The shell reacts to review momentum: open follow-up work warms it up, re-review makes it wobble, and a clean final review lets it hatch.
• Hatchability usually comes from sufficient real-behavior proof, no blocking P0/P1/P2 findings, no security attention needed, and clean correctness. A merged PR is already final, so merge makes the egg hatchable independently.
• The hatch is seeded from this repository and PR number, so the same PR keeps the same creature; the reviewed head SHA can only change safe visual details.
• Rarity is just collectible sparkle: 🥚 common, 🌱 uncommon, 💎 rare, ✨ glimmer, and 🌈 legendary.