Docs Governance

    1

    Keep project docs, decisions, and verification evidence useful across AI coding sessions.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    docs-governance

    Example session with this skill installed

    Use docs-governance to audit a synthetic project in read-only mode. It has CLAUDE.md, CLAUDE_MAP.md, PROJECT_STATUS.md, and PROJECT_LOG.md. CLAUDE.md links to docs/spec.md, but that file does not exist. Run the deterministic checker, report the complete result and its limits, recommend the smallest useful next step, and verify that the four input documents remain unchanged.

    • Read your context and instructions
    • Compiled the docs-governance

    Read-only audit of a deliberately broken sample project

    This demonstration uses a synthetic project, not customer data. The project has four small governance documents. CLAUDE.md links to docs/spec.md, which does not exist.

    The packaged checker was run after extracting the ZIP into a separate installation directory. It returned exit code 1 and correctly reported the missing specification. Hashes of all four input documents were identical before and after the audit: no project file was modified.

    Smallest useful next step

    Confirm whether a specification already exists elsewhere, then update the link to the authoritative document. If there is no specification, record that gap and establish its scope before creating one. Re-run the audit after an authorized fix.

    Complete deterministic output

    {"schema_version":1,"scope":"full","requested_base_ref":null,"base_ref":null,"head_commit":null,"worktree_dirty":null,"status":"fail","exit_code":1,"counts":{"error":0,"fail":1,"unverified":3,"warning":0,"pass":9},"findings":[{"check":"spine.carrier","status":"pass","scope":"spine","message":"CLAUDE.md 存在","path":"CLAUDE.md","line":null,"evidence":{}},{"check":"spine.carrier","status":"pass","scope":"spine","message":"CLAUDE_MAP.md 存在","path":"CLAUDE_MAP.md","line":null,"evidence":{}},{"check":"spine.carrier","status":"pass","scope":"spine","message":"PROJECT_STATUS.md 存在","path":"PROJECT_STATUS.md","line":null,"evidence":{}},{"check":"spine.carrier","status":"pass","scope":"spine","message":"PROJECT_LOG.md 存在","path":"PROJECT_LOG.md","line":null,"evidence":{}},{"check":"spine.paths","status":"pass","scope":"spine","message":"脊柱与可选载体中的路径引用存在","path":null,"line":null,"evidence":{}},{"check":"status.deleted","status":"pass","scope":"spine","message":"删除区无复活","path":"PROJECT_STATUS.md","line":null,"evidence":{}},{"check":"log.append-only","status":"unverified","scope":"spine","message":"无 Git 提交基准,历史只追加完整性未验证","path":"PROJECT_LOG.md","line":null,"evidence":{}},{"check":"log.event-count","status":"pass","scope":"spine","message":"PROJECT_LOG 活跃事件 0 条,未超过 200","path":"PROJECT_LOG.md","line":null,"evidence":{"count":0,"threshold":200}},{"check":"context.carrier","status":"unverified","scope":"context","message":"CONTEXT.md 未创建;没有稳定领域词汇时无需补空壳","path":"CONTEXT.md","line":null,"evidence":{}},{"check":"adr.index","status":"unverified","scope":"adr","message":"docs/adr/ 尚未创建;ADR 按实际决策懒创建","path":"docs/adr","line":null,"evidence":{}},{"check":"links.local","status":"fail","scope":"artifacts","message":"Markdown 链接断裂:CLAUDE.md -> docs/spec.md","path":"CLAUDE.md","line":null,"evidence":{"target":"docs/spec.md"}},{"check":"tests.references","status":"pass","scope":"artifacts","message":"未发现跨文档 TEST-ID 引用","path":null,"line":null,"evidence":{}},{"check":"docs.orphans","status":"pass","scope":"artifacts","message":"docs/ 下未发现疑似孤儿文档","path":null,"line":null,"evidence":{}}]}
    

    The output also marks unavailable Git history as unverified. Deterministic checks establish mechanical evidence; semantic review and host integration need their own checks.

    data. The project has four small governance documents. CLAUDE.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    docs-governance helps long-running AI-assisted projects keep knowledge, decisions, and verification evidence trustworthy across sessions and agents.

    Start with a read-only documentation audit: identify broken references and drift, preserve authoritative sources, and get the smallest useful next action before changing files.

    Included workflows cover living documentation, AGENTS.md / CLAUDE.md entrypoints, domain context, architecture decision records, change impact, product-document baselines, contract-first collaboration, test rationale, and module regression evidence. The ZIP includes coordinated methodologies with supporting scripts, templates, and references. Small projects can stay small.

    Free and MIT-licensed. Source, native plugin installation, examples, issues, and updates: https://github.com/qshanx/docs-governance

    Try it: “Use docs-governance to audit the current project in read-only mode and recommend the smallest useful next step.”

    Python 3.10+ is needed for deterministic checks; Git is needed for history evidence. Methodologies are primarily Chinese. The portable ZIP has been checked with a synthetic project, but automatic activation across every listed agent is not verified. Native slash commands and hooks require installation from the GitHub project. Audits report evidence and gaps; they do not automatically repair project documentation.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Permissions

    Terminal / Shell
    Read Files
    Environment Variables
    Write Files

    File Scopes

    **/*.md

    Read-only audits read project documentation and use the local shell/Python checker. Git history checks read repository metadata. Documentation updates and workflow initialization write project files only when authorized. Hooks may read project environment variables. No connected external account is required; GitHub is the public source and optional native installation destination.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    It combines deterministic documentation checks with workflows for reviewing sources of truth, project context, decisions, and verification evidence. Checks report broken local links and other mechanical issues; semantic inconsistencies need an agent review. It helps reduce outdated context but does not guarantee correct AI output.

    The methodologies can be used by agents able to read local SKILL.md files and their supporting materials. Native Claude Code and Codex installation instructions are maintained on GitHub. Automatic activation of this portable ZIP has not been verified for every agent shown by the marketplace.

    Use Python 3.10+ for the deterministic scripts and Git for history checks. Reading the methodologies does not itself require running Python. Start with a read-only audit, then authorize any desired documentation updates.

    The free MIT-licensed ZIP includes the docs-governance router, coordinated methodologies, runtime scripts, templates, and references. The full source, native plugin manifests, installation instructions, examples, issues, and latest updates are at https://github.com/qshanx/docs-governance. ZIP updates on Agensi are uploaded manually.

    It can support an ongoing coding session by routing a change to relevant context, decision, documentation, contract, or test-evidence workflows. Use only the workflows your project needs, and authorize writes when a reviewed change is ready.

    New in Technical Documentation