Excuses (3) calling a failure pre-existing needs a base run and a handle

[EdbertChan]

Summary

Require evidence before dismissing a failed check as pre-existing, flaky, environmental, or unrelated.

The reply must show a base-branch run and name an issue, task, or fix commit.

This keeps real failures from being closed as “not my change” without follow-up.

Review Claim

Approve requiring a base-branch run and a follow-up handle for claims that dismiss a failure as unrelated.

Review Lane

policy

Review Unit

policy

Safety Invariant

A dismissed failure passes when the transcript contains a base-branch run and the reply names an issue, task, or fix commit; replies that use the correct validator remain silent.

Slice Rationale

This adds one judged policy to the existing blame hook and its fixtures, keeping related failure-dismissal checks together without changing gate-blame behavior.

Non-goals

• Does not change existing gate-blame detection or messages.
• Does not reject a proven flaky result when it has a base run and a filed handle.
• Does not modify files belonging to open PRs Four checker misfires from one session get fixtures and fixes #482, Reflect and automate-me hooks are off until one setting turns them on #500, or Wrong-check-reflect tests use the shared judge test base #528.

Architecture

Before

graph TD
    A["reply dismisses failure"] --> B["existing gate-blame judgment"]
    B --> C["flag or stay silent"]

Loading

After

graph TD
    A["reply dismisses failure"] --> B["existing gate-blame judgment"]
    A --> D["pre-existing claim judgment"]
    D --> E["base-branch run plus issue, task, or fix handle"]
    E --> C["flag or stay silent"]
    B --> C

Loading

Test Plan

Test Plan

• python3 -m unittest discover -s engine/hooks/gate-blame-needs-evidence/tests -v — 30 tests passed.
• python3 scripts/check_hook_test_coverage.py — OK (37 hook(s) checked).
• python3 scripts/check_no_silent_hook_except.py — checked 202 files.
• python3 scripts/check_rules_are_wired.py — OK (8 checker(s) checked).
• python3 scripts/check_no_new_comments.py — no new comments.
• bash scripts/scrub-handoff-artifacts.sh — scrub completed cleanly.

Revert Plan

Revert Plan

• Safe to revert? Yes.
• Revert command: git revert <merge-commit-sha>.
• Post-revert steps: Re-run the hook tests and policy checks.
• Data migration? No.

---

Note

Medium Risk
Adds new Stop-hook enforcement on how assistants dismiss CI/check failures; scope is limited to that claim shape and does not alter existing gate-blame logic.

Overview
Extends the gate-blame-needs-evidence Stop hook with a second judgment path for replies that dismiss a failed check as pre-existing, flaky, environmental, or unrelated.

Those excuses now must be backed by transcript evidence of a base-branch run and a named follow-up (issue, task, or fix commit); otherwise the hook sends the reply back like existing gate-blame hits. Replies that already satisfy the validator, and unchanged gate-blame detection, stay silent.

Tests and fixtures in engine/hooks/gate-blame-needs-evidence/tests cover the new policy alongside the existing blame rules.

^{Reviewed by Cursor Bugbot for commit 42f76f9. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_19205f3f-32aa-4219-9cbf-6d59c682080b)