audit

Audits an open-autonomy installation for conformance drift, governance gaps, and doctrine violations.

Updated Sep 2, 2026
One-click install
npx skills add https://github.com/volter-ai/open-autonomy-compiler --skill audit-volter-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: audit
Source: https://github.com/volter-ai/open-autonomy-compiler/tree/main/.claude/skills/audit
Command: npx skills add https://github.com/volter-ai/open-autonomy-compiler --skill audit-volter-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? An open-autonomy install can silently drift from its own governance rules: branch protection may not match its prescription, doctrine files may reference missing resources, or local adaptations may violate core invariants. This Skill performs a read-only, nine-check conformance audit of the install itself and files a dated report without ever fixing or merging anything. ## Core Features & Use Cases - Nine-check conformance audit: Verifies direction documents, tracker preset naming, branch-protection parity, manifest integrity, doctrine-vs-reality drift, fence and loop wiring, philosophy conformance against the invariant tiers, misplaced profile content, and board health. - Profile-aware without hardcoding: Reads live facts from .open-autonomy/autonomy.yml (codeHost, capabilities, policy box) so the same skill works across simple-gh-sdlc, simple-sdlc, and self-driving installs, marking checks N/A only with cited facts. - Self-throttled cron operation: On cron-fired runs it checks for a recent report or open audit PR and stops early, while operator dispatches always get a full fresh run. - Use Case: An operator dispatches the audit agent weekly to confirm that a self-driving repo's branch protection still matches provision.json, that no agent has gained merge capabilities, and that the board's dispatch set is sane — receiving a PASS/FAIL/N/A report as a docs-only PR. ## Quick Start Dispatch the audit agent against this repository and ask it to run the full nine-check conformance audit and produce a dated report under docs/audits.

Frequently Asked Questions about audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a conformance audit on an open-autonomy install?▼

Dispatch the audit agent explicitly, for example with AUTONOMY_AGENT=audit node scripts/run-agent.mjs locally or workflow_dispatch on the audit workflow for GitHub Actions. It also runs on its own weekly cron with self-throttling if a recent report exists.

What does the open-autonomy audit skill check?▼

It runs nine checks covering direction documents, tracker preset naming, branch-protection parity, manifest integrity, doctrine reference drift, fence and loop wiring, philosophy conformance against invariant tiers, misplaced profile content, and board health via ztrack.

Does the audit agent fix the problems it finds?▼

No. The audit is strictly read-only and report-only: it files findings in a dated report under docs/audits and never edits skills, workflows, policy, or product files. Fixes are the dispatcher agent's or operator's responsibility.

Why does the audit skip some checks as N/A on certain installs?▼

Checks are marked N/A only when a cited fact makes them inapplicable, such as codeHost: local-git having no branch protection to verify, or a missing proxy_host meaning there is no egress allowlist to check. It never branches on profile names.

How does the audit handle a 404 from the GitHub branch protection API?▼

A 404 from the /protection endpoint is treated as credential masking, not absence of protection. The audit falls back to the branch endpoint's .protected flag and required_status_checks before concluding anything about live protection.

When does a cron-triggered audit run get throttled?▼

A cron-fired run stops early if a drift-mode report under docs/audits is less than about seven days old or an audit-prefixed PR or branch is already open. Operator dispatches always bypass the throttle and run fully.