A mandatory discovery system for Claude Code that enables token-efficient lazy loading of specialized agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-registry
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 agent-registry 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 Agent Registry?
Agent Registry is a sophisticated lazy-loading system designed for Claude Code agents. By replacing the traditional method of loading all agents upfront, this Openclaw Skills solution eliminates the common 16k token warning and significantly reduces context window usage by approximately 95%.
This skill acts as a centralized directory, allowing users and AI models to search for specialized expertise and load only the specific instructions required for the current task. It ensures that your development environment remains lightweight and responsive, even when managing a vast library of custom agents.
Agent Registry Use Cases
- Reducing token overhead in large-scale Claude Code projects to prevent context exhaustion.
- Discovering specialized agents based on user intent or specific project requirements dynamically.
- Managing and migrating a high volume of AI agents (300+) without hitting hard token limits.
- Automating agent selection for complex, multi-step development workflows through keyword-based intent extraction.
How Agent Registry Works
- The system extracts intent keywords from the user prompt to identify the specific expertise required for the task.
- A search command is executed to find the most relevant agents within the registry using relevance scores (0.0-1.0).
- The user or the AI reviews the results and selects the best-matching agent from the search output.
- The specific agent's instructions are fetched and loaded into the current session via the get tool, bypassing the need for pre-loading.
- The task is executed using the freshly loaded agent's specialized context and instructions.
Agent Registry Setup
Install the Skill
Install the registry directly using the CLI to integrate it into your Openclaw Skills environment:
npx skills add MaTriXy/Agent-Registry@agent-registry
Migrate Your Agents
Run the interactive migration script to index your existing agents and build the registry.json index:
cd ~/.claude/skills/agent-registry
bun bin/init.js
You can also use bun bin/init.js --move to physically migrate files into the registry instead of copying them.
Agent Registry Data Schema & Taxonomy
The skill organizes its registry through a centralized index and structured directory system:
| Component | Description | Location |
|---|---|---|
| Registry Index | registry.json containing metadata, descriptions, and relevance scores |
~/.claude/skills/agent-registry/ |
| Global Agents | Source directory for cross-project specialized agents | ~/.claude/agents/ |
| Project Agents | Source directory for project-specific overrides | .claude/agents/ |
| Search Tools | CLI utilities for list, search, and retrieval | bin/ |
Agent Registry Advanced Features
- Automatic search hooks triggered on UserPromptSubmit for seamless discovery without manual commands.
- Paged search capabilities (
search-paged.js) designed for enterprise-scale registries with over 300 agents. - Interactive migration UI with visual token indicators (green/yellow/red) to identify heavy agents before loading.
- Support for both global and project-level registry overrides for flexible environment management.
- Zero-dependency core functionality optimized for the Bun runtime environment.
SKILL.md
Loading
Related Openclaw Skills
agent-browser
A specialized headless browser CLI that uses accessibility trees and reference-based selection for reliable AI-driven web automation.
Bug Report Generator
A professional utility that transforms vague bug descriptions into structured, actionable reports with reproduction steps and severity levels.
Agent Emergence Test
A comprehensive testing framework designed to probe AI agents for emergent properties, self-awareness, and cognitive capabilities beyond their explicit training.
Pixiv
A comprehensive CLI-based skill for searching, downloading, and publishing illustrations and manga on the Pixiv platform.
Audio Reply Skill
A sophisticated text-to-speech skill that enables AI agents to read web articles aloud or respond using conversational audio synthesis.
Linear
A Node.js-based CLI skill for managing Linear issues, projects, and workflows through the official API.
METADATA
Featured*
