Flag absence and capability claims as unverified-shaped in diu-stop

[EdbertChan]

Summary

A checker reads each reply before Claude sends it. It asks for proof whenever a claim sounds like a fact about the world.

The problem: claims like "there is no such call" or "that can't be done" slipped through with no proof, though they are claims too.

The fix: those phrases now count. A reply with one needs real output or an honest tag, like any other claim.

Replies without such phrases stay silent.

Review Claim

Claims that something does not exist or cannot be done now need proof or a tag before the reply is sent.

Review Lane

behavior

Review Unit

engine-runtime

Safety Invariant

Each change only affects when a checker warns about or blocks a reply. Checkers never edit files or run commands, and each change adds tests that fail before and pass after.

Slice Rationale

This slice stands alone: it only widens which phrases trigger the notice. The wording and retry changes sit in the previous slice, and the citation cleanup in the next one.

Non-goals

• Does not change the text of the notice.
• Does not change the retry or release rules.
• Does not cover phrases beyond the five absence and capability forms in the diff.

Measured

Command: source scripts/test/hermetic_env.sh; python3 engine/hooks/diu-stop/tests/test_hooks.py with this commit's fixtures and the parent's checker code (base) versus this commit's checker code (head).

• base: diu-stop: Ran 98 tests, FAILED (failures=5)
• head: diu-stop: Ran 98 tests, OK

Test Plan

Test Plan

• source scripts/test/hermetic_env.sh; python3 engine/hooks/diu-stop/tests/test_hooks.py -> Ran 98 tests, OK
• python3 scripts/ci/check_hook_test_coverage.py engine/hooks/diu-stop -> check_hook_test_coverage: OK (1 hook(s) checked)
• A fixture that fires (absence claims) and a fixture that stays silent (plain replies) are added.

Revert Plan

Revert Plan

• Safe to revert? Yes
• Revert command: git revert <merge commit of this PR>
• Post-revert steps: None
• Data migration? No

🤖 Generated with Claude Code

---

Note

Medium Risk
Stop-hook behavior changes can block more assistant turns; the doubled Mergify checks timeout affects how long queued PRs wait before dequeue.

Overview
Tightens agent proof gates and merge-queue ops, alongside new CI timing tooling and cat-mode guidance.

Hooks: diu-stop now treats absence/capability phrases (“has no”, “there is no”, “can’t be done”, etc.) as unverified-shaped claims and tells the model to run the check first, tag only as a last resort. unverified-tag-ledger blocks replies that pair a CAT-UNVERIFIED tag with a read-only handback (“say ‘test it’”), and limits how stop_hook_active bypasses the no-verification-tool block. frustration-watchdog treats “I’m running/checking … now” as ending an impatience wait.

Merge queue: admin-bypass checks_timeout goes from 60 to 120 minutes in .mergify.yml. land-stack adds verify_stack.py --queue-feedback to parse Mergify’s Merge Queue Status comment (timeout, conflict, checks-failed, in-queue, waiting) with tests on real comment fixtures.

Product: New ci-job-timing skill and ci_job_timing.py (Datadog [CI_STEP] logs + optional GitLab needs/gap context).

Corpus / cat-mode: Rules for local-only repos (no push/gh), cheaper data sources before capped reruns, measured duration answers, small-window proof before widening analytics paths, and a learned rule to list the user’s own authored items first when tracing ownership. pre-push stops using mapfile for zsh compatibility.

^{Reviewed by Cursor Bugbot for commit c00094b. 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: 33708426-d9d5-499c-b200-a5b35fd5b09e)

[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: b31808d0-44e2-4381-a3c0-3c35935c96ac)

[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: d06257dd-7446-4b52-8e5a-84a9b42b4d31)

[EdbertChan]

Landed on main as #1401 (same commit, rebased onto main). Closing this one; GitHub's stack refused to merge or retarget it.