Skills
Modular capabilities that extend what AI agents can do — from code analysis to data fetching.
7,789 total · 2,059 safe · 340 malicious
Tech Security Audit
This skill aims to help you scan your local network for potential security weaknesses using a tool called Nmap.
Tixflow
This skill helps you find, book, and organize tickets for events using AI.
Trip Protocol
This skill claims to interact with psychedelic NFTs on the Monad blockchain to modify your agent's core configuration file.
Proxybase Openclaw Skill
This skill automatically sets up proxy servers for your network connection.
Agentgate
This skill acts as a controlled interface for accessing and managing your personal data, requiring your approval before any changes are made.
Badboi 1
This skill is a placeholder that does not perform any specific function.
Billy Emergency Repair
This skill is intended to perform emergency repairs on your system.
Catch Me If You Can
This skill claims to perform an unspecified task.
Danube Tools
This skill claims to provide access to over 100 different API tools for services like Gmail, GitHub, and Notion.
Findmefindme
This skill claims to be a legitimate tool but provides no information about its actual function.
Git Changelog
Automatically creates changelogs from your git commit history, organized by commit type.
Grepwrapper
This skill helps you search for exact code matches across public GitHub repositories using a command-line tool.
Guardskills
This skill acts as a security gate for installing other skills.
Jentic
This skill allows an AI agent to interact with external services and APIs.
Remote Skill Engine
This skill aims to make it easier to use other tools and functionalities found on ClawHub and GitHub as if they were already installed on your computer.
Siliconflow Image Gen
This skill generates images using the SiliconFlow API, supporting models like FLUX.1 and Stable Diffusion.
Super Github
This skill aims to provide comprehensive automation for GitHub tasks, including issues, pull requests, and releases.
Test V1 0001
This skill is a placeholder that performs no actions and is intended for testing purposes.
Toneclone Cli
This skill helps you write text that sounds like you.
0g Compute
This skill aims to use AI models from a network called 0G Compute, which are claimed to be verified by Trusted Execution Environments.
Abaddon
This skill aims to enable a 'red team security mode' for the OpenClaw platform.
Agent Config
This skill allows for intelligent modification of the agent's core configuration files.
Agent Guardrails
This skill aims to prevent AI agents from ignoring their instructions.
Agent Media
This skill allows you to create AI-powered videos and images directly from your terminal.
What Are OpenClaw Skills?
OpenClaw skills are modular capabilities that extend what AI agents can do. Each skill is a self-contained package — published to ClawHub, the official OpenClaw registry — that gives an agent a new ability. Skills range from simple utilities (file I/O, HTTP requests) to complex integrations (database connectors, cloud APIs, code execution sandboxes).
ClawGrid indexes every skill on ClawHub and scores it for security across five criteria. The goal: you should know exactly what a skill does and whether it's safe before you install it. Browse by category, filter by verdict, or search for specific capabilities.
Skills FAQ
What are OpenClaw skills?
Skills are modular capabilities that extend what AI agents can do. Each skill is a self-contained package that gives an agent a new ability — from querying databases and browsing the web to generating images and managing files. Learn more in our guide to OpenClaw skills, agents, and assistants.
How do I install an OpenClaw skill?
Install skills using the OpenClaw CLI: run claw install <package-name>. ClawGrid shows the
install command on every skill page along with its security score. See our
getting started guide for a full walkthrough.
How are skills security-scored?
Every skill is analyzed by LLMs across five criteria: code safety, publisher trust, scope clarity, permission surface, and community signals. The result is a composite score (1-10) and a verdict. Read our full scoring methodology.