constitution-drift-audit
Audit code, docs, tests, and registries against the Constitution, ADRs, and decision map without making broad repairs. Use when checking whether the repository is drifting away from accepted architecture.
Source facts
- Repository
- clawic/Clawix
- Last source activity
- May 15, 2026 at 12:49
- Detected SKILL.md language
- English
- Stars
- 8
- Forks
- 0
Install options
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Review the source files
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Showing SKILL.md
SKILL.md
Source instructions · Read-only preview- name
- constitution-drift-audit
- description
- Audit code, docs, tests, and registries against the Constitution, ADRs, and decision map without making broad repairs. Use when checking whether the repository is drifting away from accepted architecture.
- keywords
- ["constitution","adr","drift","audit","architecture","guardrails"]
# constitution-drift-audit
Find drift between the implemented repository and the accepted ClawJS/Clawix canon.
## Procedure
1. Start with `docs/decision-map.md`; use it to identify the canonical docs and checks for the topic.
2. Read only the relevant Constitution sections, ADRs, docs, registries, and guard scripts.
3. Prefer `claw search <topic> --json` and `claw inspect ... --json` before source reads when the CLI is available.
4. Inspect implementation, docs, tests, generated manifests, and public instructions for contradiction, missing enforcement, or stale naming.
5. Classify each finding as `confirmed_drift`, `probable_drift`, `missing_guardrail`, `doc_only_drift`, or `needs_decision`.
6. Report a small repair batch for confirmed items and leave wider changes as explicit debt.
## Constraints
- Do not rewrite architecture while auditing.
- Do not treat a private workspace preference as public canon unless it has a redacted public source.
- If Constitution, ADR, and decision map disagree, record the conflict instead of guessing.
View on GitHub