ClawHub, the public registry for OpenClaw skills and plugins, is now live at clawhub.ai. Think of it as the npm of AI agent capabilities: a versioned, searchable, community-driven library where you can find, install, and update the things your OpenClaw agent knows how to do. One command to install, one command to update, and your agent gains a new ability without writing a line of code yourself.
What a Skill Actually Is
In OpenClaw terms, a skill is a versioned bundle of files that teaches your agent how to perform a specific task. At minimum, every skill contains a SKILL.md file — a markdown document describing what the skill does, when to use it, and how to execute it. That file gets loaded into the agent's system context as structured instructions. Beyond SKILL.md, skills can include supporting scripts (shell, Python, Node), config templates, and metadata controlling when the skill loads. The key insight here is composability: a single agent with five well-chosen skills is dramatically more useful than one running blank.
Finding and Installing Skills
You can browse ClawHub directly at clawhub.ai, or search from the terminal using OpenClaw's native commands: openclaw skills search "calendar" or openclaw skills search "github pr review". The search uses vector embeddings, not keyword matching — so searching by what you want to accomplish ("email triage") surfaces relevant skills even if the skill name doesn't match exactly. Installation is one command: openclaw skills install . Skills land in your workspace's skills/ directory and are tracked in .clawhub/lock.json, giving you a package-lock-style audit trail.
Publishing Your Own Skills
This is where ClawHub gets genuinely useful. Any skill you build locally can be published to share with the community — or just backed up to the cloud. The publishing flow uses the standalone clawhub CLI (npm install -g clawhub), then clawhub login to authenticate via GitHub. Publishing requires your GitHub account be at least one week old (basic spam throttle), and skills with more than three unique reports from users get auto-hidden pending moderation. The clawhub sync --dry-run command is essential before publishing — it shows exactly what would go public so you don't accidentally leak internal skills.
Practical Patterns That Actually Work
A few things matter in practice. First, search by intent: "automate my morning standup" will surface relevant skills even without exact keyword matches. Second, inspect before you trust — every skill has a public SKILL.md page you can read first. Third, use clawhub sync as a backup strategy for your own custom skills: publish to the registry and get versioned cloud backup with one command. Fourth, pin versions in production by checking .clawhub/lock.json into version control for reproducible installs.
The Bottom Line
ClawHub isn't just a library — it's a distribution channel for agent capabilities. If you're building OpenClaw setups for clients or teams, this gives you a managed way to distribute exactly what each agent needs. The registry gets more useful as it grows, and it grows every time someone publishes something they built for themselves. The best OpenClaw setups aren't the ones with the most powerful models — they're the ones with the right skills loaded for the work at hand.