[SaulBuilds]

What

1. bootstrap.sh stages only the files it creates. Before this change, step 9 ran git add -A, so in --non-interactive mode (where every prompt auto-answers yes) any unrelated file in the adopter's tree was committed with the bootstrap. It now stages an explicit list: CONFIG.md, PRODUCT_SPEC.md, the coverage baselines, INDEX/, CLAUDE.md, the renamed Sprint 0 directory plus the placeholder's removal, and .claude/settings.json only when this run seeded it.
2. New tests: tests/test_bootstrap_staging.py runs the real bootstrap in an isolated temp copy with HOME and git config overridden. tests/test_workflow_hygiene.py checks three things:
   • every ${{ }} in a run: body is an allowlisted value;
   • every uses: line is pinned, including the - uses: list form, which the existing pin test skipped;
   • each claim-grade step validates $PR_NUMBER before using it.
3. CI: the tests/ suite wasn't run by any workflow. A self-tests job in tripwires.yml now runs it. The workflows README had drifted (merge-gate.yml was missing), which left the existing README test red on main; it now has the missing row and the corrected permission count.

Standard: met

• Red first: 30ee31f is the red test. It fails on an assertion: the planted file is committed. a6307dc is the fix.
• Revert check: re-adding git add -A turns the staging test red. Each of these turns a named hygiene test red: an inline dispatch input, an inline step output, a tag-pinned or abandoned action, or dropping one numeric check. (Under the old suite, those four, plus removing the numeric check, all stayed green.)
• Suite: python3 -m unittest discover -s tests 36 OK (was 29 passing + 1 failing on main). actionlint 1.7.7 is clean on tripwires.yml.

🤖 Generated with Claude Code

https://claude.ai/code/session_01KGMeatV8UjtCB4CTbWuAv4

[github-actions]

Claim grader (shadow mode)

Each commit / PR-text element scored against the claim-compression rubric. Shadow mode: this comment is informational; merge is not blocked.

Kind	Id	Score	Verdict	Reasons
pr_title	PR #4	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured
pr_body	PR #4	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured
commit	30ee31f4	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured
commit	a6307dc9	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured
commit	5a02e37f	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured
commit	aec167af	—	grader-unavailable	no API key in env (set ANTHROPIC_API_KEY or OPENAI_API_KEY); grader returns informational no-op until configured

[BerryManifold]

Automated Codex static review of aec167af38e473ffab8c99cd916f71ae6756093a is complete. Result: blockers; details retained privately. Details and coverage limits are recorded in the project's private review log. No tests were executed by this reviewer. Both account submissions represent the same Codex review, not independent reviews.

[Matr0xshka]

Automated Codex static review of aec167af38e473ffab8c99cd916f71ae6756093a is complete. Result: blockers; details retained privately. Details and coverage limits are recorded in the project's private review log. No tests were executed by this reviewer. Both account submissions represent the same Codex review, not independent reviews.