The Skill Router serves as the master gateway and dispatcher, automatically matching user prompts to the exact subset of capabilities needed for execution.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install skill-router-hub
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 skill-router-hub 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 Router?
The Skill Router is the core traffic controller and index for an AI coding agent's environment. Instead of flooding the LLM context with every available guideline and tool, this skill is designed to be consulted first during any interaction. It evaluates incoming user prompts to determine which specialized capabilities are required and immediately loads them on demand.
By serving as a centralized directory, this architecture ensures that relevant Openclaw Skills are dynamically loaded only when needed. This approach minimizes context bloat, reduces token usage, and guarantees that specific standards, designs, and workflow patterns are applied precisely to the developer tasks at hand.
Skill Router Use Cases
- Determining which specialized domain skills (such as API design or frontend patterns) to load for a user query
- Automatically bundling multiple relevant tools, such as combining coding standards with language-specific development skills
- Reducing LLM context window clutter by dynamically loading workspace rules only when matching terms are detected
- Generating mandatory post-execution logs to verify which agent protocols were successfully triggered
How Skill Router Works
- Initial Assessment: The system intercepts the user's prompt and routes it to the Skill Router before applying other skills.
- Rule-Based Evaluation: The router analyzes the request against established routing guidelines and the registered skill index.
- Dynamic Loading: Any matched skill is automatically injected into the active runtime context using the system's load tool.
- Combo Composition: If multiple matching conditions are met, the router loads complementary combos (e.g., coupling testing protocols with backend patterns).
- Completion Logging: The system appends a structured report to the end of the final response to document routing decisions.
Skill Router Setup
The Skill Router index can be automatically compiled using the system's built-in build scripts, or manually updated as capabilities expand.
To automatically rebuild the registry, execute the compilation script from your command line:
python scripts/generate.py
If you prefer manual configuration, you can declare matching triggers inside the dedicated index boundaries of your skill file:
<!-- SKILL_INDEX_START -->
### Category Name
| Skill | Trigger |
|--------|----------|
| `custom-skill` | Specific keyword patterns or conditions |
<!-- SKILL_INDEX_END -->
Skill Router Data Schema & Taxonomy
The router maintains an organized metadata taxonomy that structures the routing lifecycle, rules, and outcomes.
Index Schema
| Key Structure | Data Type | Purpose |
|---|---|---|
SKILL_INDEX_START / END |
Boundary Tags | Wraps the auto-generated list of all registered Openclaw Skills. |
Routing Rules |
Markdown Section | Details custom match logic and combo definitions for the model to follow. |
Usage Report |
Output Template | Defines the mandatory format for declaring loaded modules. |
Standardized Usage Report
The router guarantees observability by appending this metadata structure to the bottom of all agent outputs: ```markdown
Skill Usage Report:
- [skill-name] — [Status: Loaded / Not Triggered — Notes]
Skill Router Advanced Features
- Combinatorial Loading: Intelligently groups related skills automatically, such as loading deployment blueprints alongside observability watchdogs for cloud operations
- Automated Registry Syncing: Utilizes a Python generator script to parse your codebase and keep the active routing index in sync with zero manual upkeep
- Usage Reporting Loop: Enforces an audit trail by rendering a clear breakdown of which Openclaw Skills were activated during the session
- Custom Fallback Logic: Evaluates broad domain terminology to guarantee helper skills are loaded even when users don't name specific files or tools directly
SKILL.md
Loading
Related Openclaw Skills
MaxHub Instagram Data Assistant
A powerful data analytics and search skill designed to pull comprehensive user profiles, posts, reels, and hashtag data from Instagram using the MaxHub API.
MaxHub Bilibili Data Assistant
An advanced AI coding agent skill designed to query and analyze Bilibili videos, creators, comments, and live stream data via the MaxHub API.
MaxHub Lemon8 Data Assistant
A read-only analytical assistant designed to query, track, and analyze Lemon8 post, user, and trend data via the MaxHub API.
News Based Stock Research Skill
An educational AI agent skill designed to parse public market news, extract key catalysts, and structure balanced stock watchlists or paper-trading research notes.
learn-occam
An analytical decision-making skill designed to evaluate whether a user should learn a new topic, defer it, or acquire only the absolute minimum required to solve an immediate problem.
speak-69378f
An AI-driven operational workflow skill that leverages learned insights from evaluation videos to execute step-by-step automation sequences.
Featured*
