ClawSkills
ClawSkills: The skill dock for sharp agents.
Product Overview
ClawSkills is a specialized skill registry for AI developers, built around a versioned collection of OpenClaw skills. It packages agent capabilities into folders that can be uploaded, versioned like npm, and installed in one shot—making skills easier to browse and reuse across projects.
The platform also supports searching skills using vector-based discovery, helping developers find relevant capabilities faster. Featured skills include tools for Feishu/WhatsApp-style automation via a CLI, IoT control, AI video generation, and agent capabilities such as protocol-constrained self-evolution and knowledge management.
Key features
- Upload AgentSkills bundles and store them as versioned packages, similar to npm
- Install an entire skill folder in one shot using a package-style command
- Browse highlighted skills and “latest drops” to quickly evaluate available capabilities
- Search skills with vector-based discovery to find relevant capabilities
- Support versioning with rollback-ready installs for safer skill updates
- Curate specialized agent capabilities as packaged, shareable “OpenClaw skills”
How ClawSkills works
- 1
Upload skill bundles
An author uploads an AgentSkills bundle to the ClawSkills skill registry so it can be browsed and versioned.
- 2
Install by versioned package
A developer installs a skill folder in one shot using the package-style install command (e.g., @latest or a specific version).
- 3
Search and reuse skills
The developer uses vector-based search to find the right capability and then reuses it in their agent project.
Use cases
- An AI developer building an agent workflow wants a reusable capability (e.g., sending/searching WhatsApp history via a CLI) and installs it as a versioned skill bundle instead of rebuilding from scratch.
- A team managing multiple agent capabilities needs consistent updates and the ability to roll back when a new skill version changes behavior.
- An agent developer improves performance over time by using a self-evolution capability that analyzes runtime history and applies protocol-constrained evolution based on observed outcomes.
Who is it for?
ClawSkills is for AI developers and agent builders who maintain reusable agent capabilities and want a structured way to version, install, and search skills. It also fits teams working on agent knowledge persistence and iterative improvement workflows.