A comprehensive framework for building, importing, and registering modular capabilities to enhance AI agent performance and domain expertise.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install create-skill
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 create-skill 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 Create Skill?
The create-skill framework provides a standardized approach to extending AI agents by encapsulating specialized knowledge, multi-step workflows, and technical assets into portable packages. These packages, known as Openclaw Skills, function as modular onboarding guides that allow agents to master specific domains or business logic without bloating the global context. By following this structured creation process, developers can ensure their agents remain high-performing while gaining access to complex tool integrations and scripts.
Central to this system is the principle of progressive disclosure, which optimizes token usage and activation speed. By separating core instructions from deep-dive references and executable assets, the framework allows for a three-tier loading system. This ensures that the agent only consumes the resources necessary for the immediate task, maintaining a high degree of focus and efficiency within the Openclaw Skills ecosystem.
Create Skill Use Cases
- Building domain-specific workflows for specialized coding and development tasks
- Standardizing company-specific business logic and patterns for agent consumption
- Importing third-party skill packages from zip files or directories into an existing environment
- Automating the registration of modular capabilities using specialized CLI scripts
How Create Skill Works
- Analyze the required capability and define the primary triggers and user request patterns.
- Construct a skill directory including a mandatory SKILL.md file with YAML frontmatter and concise instructions.
- Conduct a security review for any imported skill packages to ensure compliance with ethical and legal standards.
- Execute the registration scripts to finalize the integration of the new capability into the active agent library.
Create Skill Setup
To register a new skill folder into your environment, use the dedicated registration script:
python easyclaw_register_skill.py <source_folder_path>
If you are importing a packaged skill via a zip file, use the following extraction utility:
python easyclaw_unzip_skill.py "zip_path"
Create Skill Data Schema & Taxonomy
Openclaw Skills follow a hierarchical structure designed for efficiency and clarity:
| Component | Requirement | Description |
|---|---|---|
SKILL.md |
Required | Contains name, description (YAML), and core instructions (under 200 lines) |
scripts/ |
Optional | Contains executable code and specialized automation scripts |
references/ |
Optional | On-demand documentation to keep the primary SKILL.md focused |
assets/ |
Optional | Static files and templates used during task execution |
The YAML frontmatter in SKILL.md must define the unique name and a clear description to trigger the skill appropriately.
Create Skill Advanced Features
- 200-Line Rule: Enforces brevity in primary instructions to ensure rapid activation and reduced token cost.
- Three-Tier Loading: Metadata is always available, while instructions and resources load only when the skill is triggered.
- Tiered Freedom Levels: Supports high-freedom text instructions, medium-freedom pseudocode, and low-freedom rigid scripts.
- Multi-Model Compatibility: Designed to work across different LLM backends by focusing on modularity rather than model-specific prompting.
- Automated Extraction: Built-in safety protocols for unzipping and reviewing Openclaw Skills from external sources.
SKILL.md
Loading
Related Openclaw Skills
Bot Customizer
An automated utility to personalize AI agent personas and instructions by merging custom user requirements into existing agent configurations.
Instagram Saver
An automated downloader that extracts high-resolution images and videos from Instagram posts using the Cobalt API.
Weather in China (weather-zh)
A zero-token Python script for retrieving real-time weather data and lifestyle indices for Chinese cities directly from official sources.
Subskill Generation Rule
A structural protocol for AI agents to organize code into modular subskills and centralized data directories.
Cheat Code
Cheat Code expands your AI agent's capabilities by providing real-time synthesis and data from outside its static training set.
Skill Hunter
A powerful discovery and security vetting engine designed to help developers navigate and install over 10,000 available skills on ClawHub.
Featured*
