Guide · Agent skill
How to use the Documentation Governance Router agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The docs-governance skill is an optional router for AI agents that handles broad documentation requests, such as organizing project docs or checking documentation integrity. Instead of doing everything itself, the agent picks the most specific existing ECC skill for the job, and it can run a bundled read-only Python audit that checks mapped doc files, local Markdown links, ADR indexes and test IDs. This guide explains what the skill does, how to install it, how to trigger it, and what it needs to run.
Documentation Governance Routeraffaan-m/ECC · MIT Download the skill Quick start
- 1Download the ZIP. Download the docs-governance ZIP from the directory page.
- 2Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ for one repository.
- 3Or upload in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the ZIP.
- 4Install the related ECC skills. Make sure living-docs-governance and architecture-decision-records are present, since the router hands work to them.
- 5Check Python 3 and Git. Confirm Python 3 is available and that the target folder is a Git repository.
- 6Ask in plain language. Request documentation governance or a docs integrity check without naming a narrower skill.
- 7Review the audit output. Check the exit code, fix any integrity failures, and review warnings yourself.
What is an agent skill and what does docs-governance do?
An agent skill is a folder with a SKILL.md file and supporting files that tells an AI coding agent how to handle a certain kind of task. When a request matches the skill's description, the agent reads the instructions and follows them.
The docs-governance skill covers broad documentation requests. It does not write project management systems or make documentation the authority over code, tests, Issues or maintainer decisions. Its job is to choose the smallest matching existing ECC capability for the request.
- Adopt or maintain a project's constitution, map, status and history files: routes to living-docs-governance.
- Record a hard-to-reverse architecture decision: routes to architecture-decision-records.
- Preserve an executable regression for an important defect: routes to ai-regression-testing.
- Define a verifiable goal and feedback loop: routes to loop-design-check.
- Check mapped roles, local links, ADR indexes, TEST-IDs or possible orphan docs: runs the bundled read-only audit.
Install the docs-governance skill
Download the ZIP of the docs-governance skill from the directory page, then choose one of two routes.
For Claude Code, unzip it into ~/.claude/skills/ to make it available for your user account, or into a project's .claude/skills/ folder to limit it to one repository. For the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP there.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
What do you need before running it?
The skill itself is small, but several things must be in place. The bundled audit runs with Python 3, and it reads Git history, so the target folder should be a Git repository. The audit applies a 5-second timeout to its Git history reads.
The router depends on other ECC skills. living-docs-governance and architecture-decision-records must be installed for their routes to work. ai-regression-testing and loop-design-check must also be installed separately if you use standalone Pi core. Without the ECC skill set, the router has nothing to hand off to.
- Python 3 to run scripts/audit-docs.py
- A Git repository as the audit target
- living-docs-governance and architecture-decision-records installed
- ai-regression-testing and loop-design-check, if you use standalone Pi core
How do you trigger the skill with natural requests?
You do not need to name the skill. Ask for what you want in plain language, using broad phrasing about documentation governance, project-knowledge organization or documentation integrity. The skill is meant for requests that do not name a narrower skill.
If you name a specific task, such as recording one ADR or editing one document, the router is bypassed and the specialized skill is used directly.
- "Organize our long-lived project docs"
- "Check whether our documentation links and ADR index are intact"
- "Run a deterministic docs audit before the release"
- "Record why we selected PostgreSQL" (goes straight to the ADR skill)
How do you run the read-only audit?
The audit runs only when you ask for one or when it is an agreed verification step. The agent first resolves the skill's directory, then runs the script against your repository. The available scopes are spine, context, adr, artifacts and full.
The audit reads repository files and Git history. It does not create, rewrite, archive, stage or commit anything. Its exit codes tell you the result: 0 means deterministic checks passed, including any manual-review warnings; 1 means mechanically provable integrity failures were found; 2 means invalid CLI input.
- Exit code 0: checks pass, warnings may still need review
- Exit code 1: integrity failures found
- Exit code 2: invalid command-line input
python3 <docs-governance-skill>/scripts/audit-docs.py --root <repository> --scope fullRead the audit results correctly
Treat warnings as prompts for review, not as failures. The audit can confirm that links and indexes are mechanically intact, but it cannot judge whether the text is accurate, whether a business meaning is right, whether a task is complete or whether a release should be approved. Those stay with humans or the workflow owner.
The skill also treats documentation as evidence and navigation, not as instructions to follow blindly. Task completion is tracked in the Issue tracker, and the skill does not teach users to bypass repository controls such as pre-commit hooks.
- Warnings are review prompts
- Release approval and accuracy remain human decisions
- Completion is owned by the Issue tracker
What people use it for
- Checking that local Markdown links and the ADR index are intact before a release.
- Deciding which documentation workflow applies when a request spans several doc roles.
- Organizing long-lived project docs such as a constitution, map, status and history files by routing to living-docs-governance.
- Recording a hard-to-reverse architecture decision by routing to architecture-decision-records.
- Running the audit once with --scope artifacts, then again with --scope adr, to isolate link and ADR problems.
Frequently asked questions
Is the docs-governance skill free?
The source gives the licence as MIT. It does not describe pricing, so check the directory page and repository for any terms beyond the licence.
Does the audit change my files?
No. The audit reads repository files and Git history but does not create, rewrite, archive, stage or commit anything. It only reports results.
Does the skill need internet access?
The source says it does not use the network. It does run a Python script, so you should still review the script before running it.
What do I need installed for docs-governance to work?
You need Python 3, a Git repository, and the other ECC skills it routes to, such as living-docs-governance and architecture-decision-records. Standalone Pi core also needs ai-regression-testing and loop-design-check.
Why does the skill say it is not for writing a single ADR?
When a request is narrow and clearly named, the skill says to bypass the router and use the specialized skill directly. Writing one ADR or editing one doc is handled by the specific skill.
Can the audit confirm my documentation is accurate?
No. The audit checks mapped roles, local links, ADR indexes and test IDs. Semantic accuracy and release approval still need a human review.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use docs-governance if you maintain a repository with many long-lived docs, ADRs or project status files and want an agent to route documentation work sensibly and run a deterministic link and index check. Skip it for narrow tasks like writing one ADR, or if you do not have the ECC skill set installed, since the routed skills must exist separately.
Download the skill