A comprehensive framework and toolkit for developing modular extensions that equip AI agents with specialized procedural knowledge and tool integrations.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install pro
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 pro 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 Skill Creator?
Skill Creator is a foundational meta-skill designed to guide developers through the entire lifecycle of building Openclaw Skills. These skills act as specialized onboarding guides, transforming general-purpose AI models into domain experts. By utilizing a modular architecture, Skill Creator ensures that agents remain context-efficient while gaining access to complex workflows, deterministic scripts, and rich reference materials.
The core philosophy of this tool centers on progressive disclosure and context management. It provides the necessary guardrails to create extensions that are both powerful and token-efficient, ensuring that Openclaw Skills only consume the context window when absolutely necessary for the task at hand. This approach allows developers to scale agent capabilities without overwhelming the model's memory.
Skill Creator Use Cases
- Developing custom workflows for specialized domains like legal, finance, or engineering.
- Creating reusable tool integrations for specific file formats or proprietary APIs.
- Scaling AI agent capabilities across a team by packaging company-specific knowledge and schemas.
- Automating repetitive coding tasks through deterministic script execution.
- Optimizing context window usage in complex agentic conversations.
How Skill Creator Works
- Analyze concrete usage examples to determine the required level of specificity and freedom for the agent.
- Plan the skill architecture, identifying necessary scripts, reference documents, and static assets needed for the workflow.
- Initialize the directory structure using the provided CLI tools to generate a standardized SKILL.md template with proper metadata.
- Implement the procedural logic in SKILL.md using imperative instructions and populate the resource directories with functional code and data.
- Validate and package the development folder into a distributable .skill format using automated scripts.
- Refine the skill through iterative testing based on real-world performance and agent feedback loop.
Skill Creator Setup
To begin developing Openclaw Skills, ensure you have the core toolkit available in your environment. Use the initialization script to scaffold a new project and the packaging script to finalize it:
# Initialize a new skill directory
python scripts/init_skill.py <skill-name> --path <output-directory>
# Edit your SKILL.md and add resources to scripts/, references/, or assets/
# Package and validate your skill for distribution
python scripts/package_skill.py <path/to/skill-folder>
Skill Creator Data Schema & Taxonomy
Each skill follows a strict hierarchical directory structure to ensure compatibility within the Openclaw Skills ecosystem:
| Component | Description | Context Loading |
|---|---|---|
| SKILL.md | Primary entry point with YAML frontmatter and core instructions | Loaded on trigger |
| scripts/ | Executable Python or Bash scripts for deterministic tasks | Execution only |
| references/ | Documentation and schemas loaded into context on-demand | Conditional |
| assets/ | Static files (templates, icons, boilerplates) used in final output | Output only |
Note: The YAML frontmatter must strictly contain name and description fields to drive the agent's primary triggering mechanism effectively.
Skill Creator Advanced Features
- Progressive Disclosure Architecture: Utilizes a three-level loading system (Metadata to SKILL.md to Resources) to maximize context efficiency.
- Multi-Step Workflow Logic: Supports complex sequential processes and conditional branching within the agent's procedural instructions.
- Deterministic Execution: Allows agents to run bundled scripts without loading the source code into the context window, ensuring reliability and token savings.
- Domain-Specific Partitioning: Organizes reference materials by sub-domain to prevent irrelevant context bloat during specialized queries.
- Automated Validation: Integrated packaging scripts that verify frontmatter integrity, naming conventions, and directory structure before distribution.
SKILL.md
Loading
Related Openclaw Skills
SchemaPin
A cryptographic security framework for signing and verifying AI tool schemas to prevent tampering and man-in-the-middle attacks.
AgentPin
A domain-anchored cryptographic identity protocol for AI agents and verifiable credentials integrated with Openclaw Skills.
Objective Evaluator
A sophisticated auditing skill that evaluates products and designs using the classic principles of Tony Fadell, Don Norman, and Dieter Rams.
MBTI Persona
A sophisticated configuration skill that maps the 16 Myers-Briggs personality types to your AI agent to customize communication and decision-making styles.
gog: Google Workspace CLI
A comprehensive command-line interface for managing Gmail, Calendar, Drive, Contacts, Sheets, and Docs via OAuth.
Bind Protocol MCP Server
A protocol-grade MCP server for privacy-preserving credential verification and zero-knowledge proof generation.
Featured*
