Context
PR #111 moved summary-doc discovery into the Evidence baseline checker and made the pre-Go CI classifier read a generated manifest adapter.
The RAS review loop for PR #111 ended clean under the configured gate policy, but left low/nit/info follow-ups that should not block the PR.
Follow-up work
- Add a cheap shape guard in
scripts/classify-check-changes.sh so non-comment manifest entries must match docs/*.md; fail with status 2 on malformed entries.
- Make
scripts/classify-check-changes.sh --list-summary-docs propagate summary_doc_refs failures explicitly instead of relying on set -e.
- Add shell tests for malformed manifest entries and explicit failure propagation.
- Add a short note or help text clarifying that
tools/evidencebaseline --list-summary-docs is a maintainer dry-run/inspection companion to --write-summary-docs.
Evidence
RAS review-fix run: 20260614T204231-f0307881a12cf606d27b9ee5
PR: #111
Context
PR #111 moved summary-doc discovery into the Evidence baseline checker and made the pre-Go CI classifier read a generated manifest adapter.
The RAS review loop for PR #111 ended clean under the configured gate policy, but left low/nit/info follow-ups that should not block the PR.
Follow-up work
scripts/classify-check-changes.shso non-comment manifest entries must matchdocs/*.md; fail with status 2 on malformed entries.scripts/classify-check-changes.sh --list-summary-docspropagatesummary_doc_refsfailures explicitly instead of relying onset -e.tools/evidencebaseline --list-summary-docsis a maintainer dry-run/inspection companion to--write-summary-docs.Evidence
RAS review-fix run:
20260614T204231-f0307881a12cf606d27b9ee5PR: #111