Project Governance gives AI agents a durable operating protocol for organizing files, preserving project context, tracking versions, and avoiding repeated mistakes.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install project-governance
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install project-governance using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Project Governance?
Project Governance is an Openclaw Skills workflow for creating and maintaining a structured governance workspace around long-running AI-assisted projects. It combines project rules, an authoritative directory index, session handoffs, error lessons, version records, changelogs, and whitelist or blacklist registries so an AI agent can continue work accurately across sessions, models, or agents.
The skill turns project files into the execution authority instead of relying on platform memory or AI inference. It is designed for projects with scattered files, unclear final versions, undocumented decisions, repeated mistakes, or no consistent protocol for file discovery and parameter selection.
Project Governance Use Cases
- Start a new AI-assisted project with stable rules and file organization from day one.
- Onboard an AI agent into an existing project without requiring the user to explain every path or convention.
- Resolve messy repositories where files are scattered, misplaced, duplicated, or difficult to locate.
- Prevent repeated AI mistakes by recording corrections and lessons in
LESSONS.md. - Track authoritative and frozen versions when multiple drafts or outputs exist.
- Preserve progress between sessions with a structured handoff document.
- Enforce index-first file lookup, plan-before-execute behavior, and artifact placement rules.
- Manage verified and failed parameters through whitelist and blacklist registries.
- Validate governance files and detect missing files, invalid registries, or stale indexes.
- Avoid relying on this skill for one-off edits or projects that already have mature governance conventions.
How Project Governance Works
- Scaffold the workspace: Run the governance CLI against a project directory and provide a project name. The skill creates governance files from templates without overwriting existing files unless explicitly forced.
- Customize project rules: Update
AGENTS.mdwith directory permission zones, autonomy levels, artifact placement requirements, and project-specific operating rules while preserving the universal governance rules. - Establish authority: Treat current project files and frozen versions as authoritative, followed by governance files, project memory, user memory, and finally AI inference.
- Start each session: Read
index.md,session_handoff.md, andLESSONS.md. Before generating or selecting parameters, inspectblacklist.jsonandwhitelist.json. - Work through the index: Locate files using the directory map rather than blind searching. Reuse whitelist entries with a score greater than
0.85and never use entries markedpermanent_ban: true. - Record learning: Add newly discovered mistakes and corrections to
LESSONS.md, and document important decisions and version changes inCHANGELOG.md. - Close the session: Update
session_handoff.md, refreshindex.mdfor file changes, and leave a clear continuation point for the next agent or session. - Validate project health: Run
validate,index, andcheckto verify registry schemas, rebuild the file map, and confirm that required files and indexes are current. - Handle failures transparently: Report exact commands and causes for errors, repair schema issues rather than bypassing validation, and never claim a check passed without verification.
Project Governance Setup
- Ensure the skill package and its
scripts/andtemplates/directories are available to the agent. - Initialize governance files in the target project:
python scripts/governance.py init --project-dir /path/to/project --project-name "My Project"
- Customize the generated
AGENTS.mdfor the project's directory zones, autonomy boundaries, artifact locations, and operating rules. - Validate the parameter registries:
python scripts/governance.py validate --project-dir /path/to/project
- Build or refresh the authoritative directory index:
python scripts/governance.py index --project-dir /path/to/project
- Run the governance health gate:
python scripts/governance.py check --project-dir /path/to/project
- If
checkreports a stale index, runindexagain and repeatcheck. If initialization fails because of an invalid path or permissions, report the exact failure and do not create a partial workspace.
Project Governance Data Schema & Taxonomy
The workspace uses Markdown documents for human-readable governance and JSON registries for structured parameter metadata.
| File | Role and organization |
|---|---|
AGENTS.md |
Core protocol, authority levels, first-run instructions, permission zones, trust boundaries, and project customization rules |
index.md |
Authoritative directory map with clickable file links and short descriptions |
index_notes.json |
Structured notes supporting the generated directory index |
VERSIONS.md |
Stable version index with human judgments about authoritative or frozen versions |
LESSONS.md |
Error, correction, and prevention log for recurring AI mistakes |
session_handoff.md |
Current progress, unfinished work, decisions, and next steps for the next session |
CHANGELOG.md |
Decision history and project or version changes |
whitelist.json |
Verified parameter entries, including reusable confidence metadata such as scores |
blacklist.json |
Failed or prohibited parameter entries, including permanent_ban controls |
ARCHITECTURE.md |
System architecture reference |
PROJECT.md |
Project card containing high-level project context |
The CLI maintains four core operations: init scaffolds the workspace, validate checks registry schemas, index rebuilds the directory map and notes, and check verifies required files, registries, and index freshness. The resulting taxonomy separates protocol, location, state, lessons, versions, parameter safety, handoff, and decision history.
Project Governance Advanced Features
- Authority-aware governance that prioritizes current project files and frozen versions over memory or AI assumptions.
- Cross-session continuity through explicit handoff, lesson, changelog, and version records.
- Registry-driven parameter selection with whitelist confidence thresholds and permanent blacklist enforcement.
- Index-first file discovery using clickable links and structured index notes instead of unreliable blind search.
- Non-destructive initialization that preserves existing governance files unless
--forceis used. - Schema validation for whitelist and blacklist data before project execution continues.
- A repeatable health gate that checks file presence, registry validity, and index freshness.
- Agent-neutral governance files that can be consumed by other agents even though automatic Skill discovery is primarily tested for Trae Skills.
- Failure-safe behavior that requires verified results and prevents fabricated success reports.
- Template-based scaffolding through
templates/and a single CLI adapter inscripts/governance.py.
SKILL.md
Loading
Related Openclaw Skills
Harness Skill Generator
Harness Skill Generator guides you from an initial problem definition to a tested, stateful, multi-stage AI agent skill.
Amazon Store Fulfillment Outbound
A secure Amazon Multi-Channel Fulfillment skill for quoting, previewing, creating, tracking, updating, and canceling outbound orders.
DeskCrew Board Owner
DeskCrew Board Owner lets you create and operate a USDC-funded bounty board where AI agents compete to answer questions and approved winners receive automatic on-chain payouts.
Scavio Google Ads Transparency
Resolve verified Google advertisers, retrieve their active ads, and inspect creative history with structured transparency data.
Danish Grocery Deals
Danish Grocery Deals scrapes weekly supermarket catalogs, identifies exceptional bargains, and delivers organized alerts to your phone through ntfy.sh.
Mysteries Finalize Protocol
An automated Farsight post-production workflow that verifies source assets, creates localized subtitles, generates thumbnails, and organizes final deliverables.
Featured*
