Scheduled hot intake re-reviews unchanged PRs after owned comment timestamps outrun sync metadata

[PollyBot13]

What Problem This Solves

ClawSweeper repeatedly performs full scheduled-hot reviews of unchanged pull requests after updating its own durable review comment. The owned mutation advances GitHub item updated_at slightly after review_comment_synced_at, so the scheduler classifies the bot-generated timestamp as fresh target activity once the one-hour exact-review suppression window expires.

Why This Change Was Made

Live reproduction on openclaw/clickclack#117:

• Exact scheduled-hot run: https://github.com/openclaw/clawsweeper/actions/runs/30704608131
• Reviewed unchanged head d26c11abba773132c51defb6e647db3a69610003
• Queue source action: scheduled_hot_intake
• Source updated_at: 2026-08-01T12:44:07Z, created by the prior ClawSweeper review publication
• Review completed at 2026-08-01T14:52:41Z
• ClawSweeper synced its durable review comment at 2026-08-01T14:53:28Z
• GitHub then exposed the PR updated_at as 2026-08-01T14:53:29Z
• The run performed a full model review with cache_hits=0 despite unchanged head/source/pull/review activity

The policy currently compares GitHub item updated_at to local sync wall-clock timestamps. The server-visible item timestamp can land one or more seconds later, leaving a permanent false activity marker.

User Impact

Affected items generate recurring review emails, consume model/Actions capacity, churn rating/status labels, and obscure genuine review changes. #117 accumulated repeated no-finding reviews on the same head; openclaw/openclaw#114160 and other tracked items show the same pattern.

Evidence

Expected: ClawSweeper-owned comment or label mutations remain identifiable as owned activity, while human comments, pushes, reviews, pull-state changes, and explicit review commands still trigger fresh review.

Proposed fix: persist the authoritative post-mutation GitHub item updated_at and use exact equality as owned-activity coverage. Preserve fail-open behavior when that observation is missing and preserve daily/normal cadence.

AI-assisted: yes.

[clawsweeper]

Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve. Reviewed August 3, 2026, 1:34 PM ET / 17:34 UTC.

Summary
This issue is still necessary: current main deliberately classifies an item timestamp one to five seconds after ClawSweeper’s own sync as fresh activity, matching the reported repeat-review path. The focused repair is already owned by the open linked PR #1007, so this issue should remain open until that branch is rebased, reviewed against current main, and merged or replaced.

Reproducibility: yes. —current source and a focused current-main test reproduce the decision: timestamps one to five seconds after a ClawSweeper-owned sync remain eligible for review. A live GitHub run is not needed to establish the narrow scheduler defect.

Next step
An open linked PR already owns the narrow repair, so starting another ClawSweeper fix PR would duplicate the implementation effort.

Review details

Best possible solution:

Land a rebased receipt-based ownership check that suppresses only bot activity covered by a complete structural snapshot while continuing to fail open for human, incomplete, or ambiguous activity.

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

Yes—current source and a focused current-main test reproduce the decision: timestamps one to five seconds after a ClawSweeper-owned sync remain eligible for review. A live GitHub run is not needed to establish the narrow scheduler defect.

Is this the best way to solve the issue?

Yes, the linked PR’s structural-receipt direction is safer than timestamp equality because it can distinguish owned mutation churn while failing open when target-side activity is incomplete or changes the reviewed input.

AGENTS.md: found and applied where relevant.

Remaining risk / open question:

• The linked PR fix(review): suppress owned activity churn #1007 is reported dirty against an older base, so its receipt-based suppression must be rebased and re-reviewed against current main before it can safely change scheduler behavior.

Codex review notes: model internal, reasoning high; reviewed against 4f67ca516da0.

Label changes

Label justifications:

• P2: Repeated reviews consume Actions and model capacity and churn status notifications, but the issue has a bounded automation blast radius.
• impact:other: The concrete impact is review-queue, notification, and capacity churn rather than data, delivery, session, provider, or crash behavior.

Evidence reviewed

What I checked:

• Current scheduler still treats delayed owned timestamps as activity: hasActivitySinceReview suppresses timestamps only through the latest local comment or label sync time, then returns true for a later item timestamp; a GitHub timestamp that arrives after the owned mutation therefore remains fresh activity. (src/scheduler-policy.ts:97, 4f67ca516da0)
• Current regression test source-reproduces the reported timing gap: The scheduler test explicitly expects an item updated one through five seconds after either owned sync timestamp to remain due, which is the mechanism described in this issue. (test/scheduler-policy.test.ts:171, 4f67ca516da0)
• The affected decision feeds queue planning: Review planning calls shouldReviewItem, reviewPriority, and schedulerBucket, so the timestamp decision can promote an unchanged item into scheduled review work. (src/clawsweeper-review-planning-inventory.ts:57, 4f67ca516da0)
• Related feature history: The current scheduler path’s recent history includes the earlier exact-hot-intake suppression work, indicating that the current behavior evolved from this same scheduling surface rather than an unrelated subsystem. (src/scheduler-policy.ts:84, 0ced0b5e515b)
• Open implementation candidate already owns the repair: The provided GitHub context shows fix(review): suppress owned activity churn #1007 remains open and explicitly closes this issue; its structural-receipt approach is intended to preserve fail-open handling for ambiguous same-second target activity. (84160ed710bb)

Likely related people:

• Martin Cleary: Authored the recent exact hot-intake duplicate-suppression commit on the scheduler and structural-cache area. (role: introduced related scheduler behavior; confidence: high; commits: 0ced0b5e515b; files: src/scheduler-policy.ts, src/review-structural-cache.ts)
• Peter Steinberger: Recent commits modify durable review-comment synchronization and review-orchestration behavior adjacent to the affected scheduler metadata path. (role: recent area contributor; confidence: medium; commits: e55de43cfc03, 922ebc673e12; files: src/clawsweeper-record-metadata.ts, src/scheduler-policy.ts)

How this review workflow works

• 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.