repo-audit-refactor-optimizelisted

# Repo Audit Refactor Optimize ## Overview Run a deterministic pipeline from bootstrap to verified completion. Start with capability discovery, then diagnosis, then execution, then verification. Load the reference docs on demand: - `references/bootstrap.md` - bootstrap policy, dependency state, overrides - `references/pipeline.md` - stage order, artifacts, run report - `references/activation-matrix.md` - preferred/fallback/manual/blocked behavior - `references/prioritization.md` - backlog ranking and batching - `references/verification.md` - evidence and rerun standards - `references/remediation-playbook.md` - execution discipline ## Stage Order 0. Bootstrap. 1. Discovery. 2. Diagnosis. 3. Synthesis. 4. Execution. 5. Verification. 6. Run report write. ## Stage 0: Bootstrap Run the checker first and only move forward on green or safe degraded mode: ```bash python3 scripts/check_skill_requirements.py \ --repo /path/to/target-repo \ --out-dir /tmp/repo-audit-refactor-optimize/<repo-name>/<timestamp> ``` The checker is deterministic and non-mutating. It reads the manifest and writes `bootstrap/bootstrap_report.json`, `bootstrap/bootstrap_report.md`, and `bootstrap/install_plan.md`. Rules: - Continue when all blocking lanes are usable. - Continue degraded when only non-blocking lanes are missing. - Keep bootstrap installs explicit by user approval. - Prefer `skill-installer` if available; otherwise use `npx skills add/find`. - Never auto-install local/private skill