fix: warn on too-short sha= marker in Minimize PASS review comment step - #104
Merged
twistedmelonman merged 3 commits intoAug 8, 2026
Merged
Conversation
The "Check review verdict" step already warns and logs when a skip-claude-review sha= marker is too short to trust (<7 chars) and proceeds with review. The "Minimize PASS review comment" step had no equivalent branch for the same too-short case — it fell through silently to the minimize logic with no log message at all, creating a behavioral asymmetry between the two steps that made them harder to reason about together when debugging. Adds the missing elif branch: a too-short MARKER_SHA now logs the same ::warning:: as the verdict step, then falls through to the minimize logic (matching the verdict step's "reject and proceed" behavior) rather than silently doing nothing. Verified with a standalone harness exercising all five marker states: no marker, unscoped marker, too-short scoped marker (now warns and falls through), scoped marker matching head (skips, unchanged), and scoped marker not matching head (falls through silently, unchanged — this is a distinct case from too-short and correctly has no warning, matching the verdict step's precedent of only warning on the too-short/invalid case). Committed with SKIP=zizmor per the documented escape hatch (dotfiles pre-commit/config.yaml, PR #155): zizmor lints the whole file and surfaces pre-existing findings (template-injection, known-vulnerable- actions, ref-version-mismatch, artipacked) unrelated to this diff. Confirmed identical finding count/type on main before this change via `zizmor .github/workflows/claude-blocking-review.yml` — this diff introduces zero new findings. Given this repo's wide fleet-wide blast radius and the explicit ask to keep fixes narrowly scoped/separable, those pre-existing findings are out of scope here and warrant their own dedicated triage rather than a drive-by fix or suppression bundled into this PR. Closes #92.
twistedmelonman
deleted the
claude/fix-minimize-pass-short-sha-warning-92-session1
branch
August 8, 2026 01:08
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
1 participant
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.
Summary
[skip-claude-review sha=...]marker'ssha=value is too short (<7 chars) to trust, then proceeds with review.elifbranch: a too-shortMARKER_SHAnow logs the same::warning::as the verdict step, then falls through to the minimize logic (matching the verdict step's "reject and proceed" behavior) instead of silently doing nothing.Test plan
yamllint— clean (pre-existing line-length warnings unrelated to this diff)SKIP=zizmorper the documented escape hatch (dotfilespre-commit/config.yaml, PRCLAUDE_OUTCOME_RETRYcomment slightly mischaracterizes the skipped-first-step case #155): zizmor lints the whole file and this file carries pre-existing findings (template-injection,known-vulnerable-actions,ref-version-mismatchx2,artipacked) unrelated to this diff. Confirmed identical finding count/type present onmainbefore this change — this diff introduces zero new findings. Given this repo's wide fleet-wide blast radius and the ask to keep fixes narrowly scoped, those pre-existing findings are out of scope here and warrant their own dedicated triage.Closes #92.
https://claude.ai/code/session_019yrvtEbtQxBxDmZ4Fu9GrU