feat(setup-aiengineering): self-audit + versioned re-run upgrade

[jsifalda]

Summary

• Closes the reactive-patch loop that produced six follow-up PRs to this skill: it can now detect its own gaps and propagate fixes through re-runs instead of one PR per gap.
• Adds references/baseline-checklist.md (canonical concerns + skill version) and a Step 8b coverage self-audit that surfaces any uncovered concern as skill output.
• Versioned provenance note + Step 1 re-run upgrade mode: older setups are offered newer concerns and drifted blocks are refreshed (diff-and-ask, never clobber).
• Fixes three silent-failure gaps: delegated skills guarded for availability (Step 6), worktree module host-gated on Claude Code (Step 7), detected commands confirmed before injection (Step 5).

Summary by CodeRabbit

• New Features
  • Added versioned upgrade mode for existing AI-engineering setups, with newer baseline concerns surfaced for review.
  • Added a baseline checklist and end-stage coverage audit showing installed, skipped, or uncovered items.
  • Added safer refresh behavior that preserves local edits and avoids overwriting customized content.
• Bug Fixes
  • Improved setup reliability by validating detected commands, guarding unavailable capabilities, and clearly labeling unsupported host-specific steps.
  • Added version tracking and drift handling for generated verification guidance.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 83e1c22e-7791-41f1-903f-a5d95f83ea78

📥 Commits

Reviewing files that changed from the base of the PR and between 2d5f6ac and 1e99e02.

📒 Files selected for processing (5)

• README.md
• changelog/20260713192819-setup-aiengineering-selfaudit-versioning.md
• skills/setup-aiengineering/SKILL.md
• skills/setup-aiengineering/references/baseline-checklist.md
• skills/setup-aiengineering/references/verification-protocol.md

---

📝 Walkthrough

Walkthrough

The setup-aiengineering skill now supports versioned rerun upgrades, validated command injection, guarded optional modules, and a mandatory baseline coverage self-audit. A canonical checklist and version/drift rules document these behaviors.

Changes

Setup skill lifecycle

Layer / File(s)	Summary
Baseline checklist and upgrade mode skills/setup-aiengineering/SKILL.md, skills/setup-aiengineering/references/baseline-checklist.md	Defines the versioned baseline and uses provenance versions to identify and offer rerun upgrades.
Validated policy injection skills/setup-aiengineering/SKILL.md, skills/setup-aiengineering/references/verification-protocol.md	Requires command confirmation, restricts dry-runs, and updates provenance and injected blocks without duplicate notes or clobbering edits.
Guarded delegated modules skills/setup-aiengineering/SKILL.md	Adds availability checks for delegated skills and restricts worktree bootstrapping to Claude Code hosts.
End-stage coverage reporting skills/setup-aiengineering/SKILL.md	Adds expanded post-check reporting and a required audit classifying every baseline concern as installed, skipped, or not covered.
Change records and index alignment changelog/*.md, README.md	Documents the new setup behavior and rewrites the unchanged skill-table row in place.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant Operator
  participant SetupSkill
  participant TargetProject
  participant BaselineChecklist
  Operator->>SetupSkill: run setup or rerun
  SetupSkill->>TargetProject: inspect provenance and detected commands
  SetupSkill->>BaselineChecklist: read versioned baseline concerns
  BaselineChecklist-->>SetupSkill: return upgrade and audit criteria
  SetupSkill->>Operator: request command confirmation and report skips
  SetupSkill->>TargetProject: apply confirmed changes and refresh drifted blocks
  SetupSkill->>Operator: report coverage audit

Loading

Possibly related PRs

• jsifalda/ai-project-rules#63: Overlaps with the Step 7 worktree auto-bootstrap behavior.
• jsifalda/ai-project-rules#75: Modifies the same verification-protocol injection area.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/setup-aiengineering-selfaudit-versioning

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}