OpenClaw Skills: How to Find, Install, and Build Your Own (2026)

What Are OpenClaw Skills?

Skills are the primary way to extend what your OpenClaw agent can do. A skill is a folder containing a plain Markdown file called SKILL.md. That file has YAML frontmatter describing the skill's name, version, and requirements, followed by a body with step-by-step instructions that load into the agent's context when the skill is active.

When you send a message that matches a skill's use case, the agent reads the instructions from SKILL.md and follows them to complete the task. Skills can also include supporting scripts, configuration files, API integrations, and any other resources their instructions reference.

The community-maintained registry ClawHub contains over 5,700 skills as of early 2026, covering almost every automation use case imaginable.

How to Install OpenClaw Skills

Method 1: The clawhub CLI

The cleanest way to install skills is via the clawhub command-line tool:

clawhub install <slug>

Other useful commands:

clawhub list # Show all installed skills clawhub uninstall <slug> # Remove a skill clawhub update --all # Update all skills to latest versions

To find the right slug, browse github.com/openclaw/clawhub and look at the folder names, each folder name is the skill's slug.

Method 2: Paste a GitHub Link in Chat

If you find a skill on GitHub that is not yet in ClawHub, paste its repository URL directly into your conversation with the agent:

Install this skill: https://github.com/chainstacklabs/polyclaw

OpenClaw recognizes the pattern, downloads the skill, and confirms installation. This works for any public GitHub repository containing a valid SKILL.md at its root.

What Skills Are Available?

ClawHub organizes skills into broad categories. Here are the most popular ones in 2026:

Productivity

• Email management and triage (archive, summarize, draft replies)
• Calendar scheduling, reminders, and meeting summaries
• Document drafting and editing
• News monitoring and daily digest creation

Development

• Code review and refactoring suggestions
• Deployment to cloud providers (AWS, GCP, Railway, Vercel)
• Git workflow automation (PRs, branch management, changelogs)
• Test generation from existing code

Finance and Trading

• Crypto wallet management and balance tracking
• DeFi operations: liquidity pools, staking, yield farming
• Polymarket prediction market trading via the polyclaw skill
• Portfolio tracking with price alerts

Web and Research

• Web scraping and content summarization
• Competitor tracking and monitoring
• SEO analysis and keyword research
• Social media monitoring

Security

• Vulnerability scanning via the clawsec suite
• Secret and credential detection in codebases
• Dependency audit and CVE monitoring

For a curated list of the best skills across all categories, see the awesome-openclaw-skills repository.

Read the trading guide for a detailed look at the finance and crypto skills specifically.

SOUL.md: Your Agent's Personality File

SOUL.md is a special file that lives in your OpenClaw memory directory at ~/clawd/. It works alongside skills but serves a fundamentally different purpose.

While skills add new capabilities, SOUL.md defines your agent's core identity:

• Communication style (formal, casual, concise, detailed)
• Values and ethical rules ("never delete files without explicit confirmation")
• Preferred tools and approaches for common tasks
• Hard rules the agent should always or never follow

The SOUL.md is loaded at the start of every reasoning cycle, before any skill instructions are processed. It acts as the agent's constitution. Skills must operate within the boundaries SOUL.md defines.

OpenClaw provides a SOUL.md template to help you get started. Customizing it is one of the highest-leverage things you can do to make the agent behave exactly how you want.

How to Write Your Own OpenClaw Skill

Building a custom skill requires only a folder and a Markdown file:

my-skill/ SKILL.md helper-script.sh (optional)

A minimal SKILL.md:

--- name: My Custom Skill description: Explain what this skill does in one sentence version: 1.0.0 requirements: - python3 --- ## Instructions When the user asks to [trigger phrase or condition], do the following: 1. [First step with precise instructions] 2. [Second step] 3. Confirm completion to the user with a brief summary. ## Rules - Never proceed without confirming the action first. - If a required input is missing, ask for it before starting.

Tips for writing effective skills:

Treat SKILL.md like a recipe for a very literal cook. The LLM executes what it reads, so ambiguous instructions produce inconsistent behavior. Be specific about triggers, steps, and edge cases.

Test your skill by installing it and running through several scenarios before sharing it. Include a "Rules" section for constraints you always want enforced.

Once your skill works reliably, submit it to ClawHub by forking github.com/openclaw/clawhub, adding your skill folder, and opening a pull request.

Frequently Asked Questions

How many skills can I install?

There is no hard limit. In practice, having too many active skills increases response time because more context loads per request. Use clawhub list to audit your skills and uninstall ones you do not use.

Do I need to restart OpenClaw after installing a skill?

No. Skills load dynamically. Newly installed skills are available in the next conversation turn without restarting the gateway.

Where are skills stored locally?

In ~/.openclaw/skills/ by default. You can inspect and edit skill files directly in that directory.

Can skills access my file system and run code?

Yes, if the skill's instructions include those operations and your OpenClaw configuration allows them. You can restrict tool access in your configuration to limit what skills can do.

What is the difference between a skill and SOUL.md?

Skills add new capabilities and are installed on demand. SOUL.md defines the agent's baseline personality and rules and is always active. Think of SOUL.md as the agent's core character and skills as learned behaviors that build on top of it.