Repository navigation
Conversation
Owner
Author
|
Superseded by the gate-integrity PR, which carries the range sentence in the coordinator's prompt (the only pipeline copy now) and in the cross-review skill's template. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
_recovery_review_promptnow tells the out-of-band reviewer that the change under review is the whole pull request as it stands at the pinned commit, naming the range (git diff origin/main...<sha>), not only the head commit's own diff. The self-test asserts the range reaches the prompt text.Why
On muesli PR #769 (2026-09-18) the recovery review returned PASS after reading one test file. The prompt said "You are reviewing EXACTLY commit ", which exists to pin the tree (#66) but says nothing about how much of it to read. Codex's own Verification section said "Exact commit … reviewed"; the head commit was an 8-line test fix on a 38-file, 4,171-line PR. The run then posted
bircher/cross-review=successand the repository'sreview-gatewent green on a PR whose substance no second vendor had read.On muesli #766 the same words produced a whole-PR review, so this is reviewer interpretation of an ambiguous prompt. The gate cannot distinguish the two readings; only the prompt can.
Evidence
084bd039…reviewed."git diff --stat origin/main...084bd03: 38 files, 4,171 insertions.git show --stat 084bd03: 1 file, 8 insertions.docs/design/front-half-live-log.md, section "#769 recovered: codex PASS on eight lines of a four-thousand-line PR" (branchdocs/live-proof-e10, PR docs: live log E10 -- the first attempt, outcome 2 #100).Verification
bash -n batch/run-queue.shpasses; the self-test's new assertion fails on the previous prompt text and passes on this one.batch/update-bundle.sh fix/recovery-review-range(which runs the self-test) and exercised by re-running--recover-pr i764 769 codex; the outcome is recorded on muesli PR #769.