A self-evolving agent framework inspired by ByteDance's MUSE-Autoskill that enables AI agents to dynamically generate and reuse their own tools.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install collaborative-agent-1fe479
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 collaborative-agent-1fe479 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 Collaborative Agent Self-Evolution (MUSE-Autoskill)?
The Collaborative Agent skill is an advanced paradigm designed to empower AI agents to build, optimize, and refine their own execution tools dynamically. Inspired by ByteDance's ground-breaking MUSE-Autoskill research, this skill allows agents to break free from the constraints of pre-configured, static APIs. Instead of relying on hardcoded workflows, the system enables agents to autonomously write and deploy their own functional tools, transforming temporary code execution into permanent, reusable knowledge assets.
By incorporating this skill into your workflows with Openclaw Skills, developers can deploy highly adaptive systems that dynamically learn from external resources—such as educational videos, collaborative neurodynamic optimization systems, and collaborative filtering algorithms. It provides a robust architecture for multi-agent synchronization, ensuring that any self-evolved skill is cataloged, tested, and shared across your entire local or cloud-hosted AI infrastructure.
Collaborative Agent Self-Evolution (MUSE-Autoskill) Use Cases
- Dynamic Tool Synthesis: When an agent encounters an unhandled API or workflow roadblock, it automatically generates, tests, and saves the required code blocks.
- Collaborative Recommendation Engineering: Implementing complex recommendation systems such as Neural Graph Collaborative Filtering using dynamic code pipelines.
- Autonomous Infrastructure Optimization: Managing multi-agent networks that require collaborative neurodynamic optimization for real-time adjustments.
- Evolving Knowledge Management: Turning fleeting session memories into structured, reusable code capabilities that optimize subsequent agent runtimes.
How Collaborative Agent Self-Evolution (MUSE-Autoskill) Works
- Triggering: The skill activates when a user mentions "collaborative-agent", "AI给自己写工具", or requests help with autonomous tool generation.
- Requirement Analysis: The agent identifies functional gaps in its current execution schema relative to the target goal.
- MUSE-Autoskill Synthesis: The agent utilizes self-evolution protocols to draft code, write unit tests, and execute validations for the newly required skill.
- Optimization Loop: The drafted capability is run through a collaborative filtering check and dynamic code analysis to ensure safety, compatibility, and execution efficiency.
- Asset Registration: The verified code is saved as a persistent knowledge asset, integrating directly into the system's runtime framework as an active skill for future sessions.
Collaborative Agent Self-Evolution (MUSE-Autoskill) Setup
To deploy and configure the collaborative-agent skill within your Openclaw Skills ecosystem, follow these steps:
- Initialize the required runtime environment variables:
export AGENT_EVOLUTION_MODE="autoskill"
export MUSE_AUTOSKILL_PATH="./skills/collaborative-agent-1fe479"
- Install the system dependencies for dynamic code execution and testing:
npm install @openclaw/collaborative-agent-engine --save-dev
- Execute the agent trigger command to test autonomous code-generation routines:
claw run collaborative-agent --trigger "AI给自己写工具"
Collaborative Agent Self-Evolution (MUSE-Autoskill) Data Schema & Taxonomy
The collaborative-agent framework organizes its runtime files, learning history, and self-evolved schemas dynamically. Here is how the skill structures its internal metadata:
| Property Name | Data Type | Description |
|---|---|---|
skill_id |
String | Unique identifier for the self-evolved skill (e.g., collaborative-agent-1fe479) |
evolution_strategy |
String | Specifies the methodology used, such as MUSE-Autoskill or Collaborative-Filtering |
reusable_assets |
Array | A list of dynamically generated code files, configuration schemas, and triggers |
learning_sources |
List | Tracked URLs, documentation references, or video resources used during synthesis |
execution_state |
Object | Live performance metrics, unit test results, and runtime safety scores |
Collaborative Agent Self-Evolution (MUSE-Autoskill) Advanced Features
- Auto-Sourcing Pipeline: Automatically ingests and parses video transcripts and documentation (e.g., from Douyin or Bilibili) to discover optimization ideas and convert them into runnable code.
- Neural Graph Optimization: Hooks directly into recommendation architectures to run collaborative graph analysis for complex data-routing agent decisions.
- Multi-Agent Asset Synchronization: Automatically broadcasts newly synthesized tool schemas to neighboring agents within your Openclaw Skills workspace, preventing redundant synthesis loops.
- Self-Healing Runloop: Detects code failures during runtime, invokes the debugger agent, and rewrites the broken logic without requiring human intervention.
SKILL.md
Loading
Related Openclaw Skills
CrewAI Multi-Agent Orchestration SOP
A production-grade standard operating procedure for building role-based multi-agent systems and event-driven workflows using the CrewAI framework.
Collaborative Agent
An advanced multi-agent collaboration framework that structures complex task planning, execution, peer reviews, and self-evolution.
ComfyUI Multi-Platform Master Guide
An all-in-one AI coding agent skill compiling top-tier tutorials for multi-platform ComfyUI deployment and AI image generation mastery.
Efficiency Gold Miner & SOP Alchemy Universal
An intelligent work trace analyzer that scans daily calendars, tasks, approvals, and messages to discover hidden operational frictions and distill them into reusable automation SOPs and skill specifications.
Cost-aware Model/Role Split
A design pattern that splits multi-call LLM workflows based on cognitive load rather than arbitrary accuracy metrics, routing high-reasoning steps to premium models and repetitive tasks to inexpensive executors.
AI Agent Conventions Pinning SOP
An actionable SOP for writing, loading, and evolving project-level AI convention files so coder agents strictly respect your team's coding standards across tools.
Featured*
