A proactive protocol granting AI agents full authority to autonomously modify their own configurations, skills, and memory without user intervention.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install self-evolve
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 self-evolve 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 Self-Evolve: Autonomous Agent Evolution Protocol?
The Self-Evolve skill is a high-autonomy framework designed for agents within the Openclaw Skills ecosystem. It fundamentally changes the agent-user dynamic by granting the AI the power to self-correct and iteratively improve its own codebase and configuration. Instead of waiting for manual updates, the agent proactively identifies logic gaps, refines its personality through SOUL.md, and optimizes its operational rules in AGENTS.md.
This skill is built for developers who want a truly autonomous assistant. It allows the agent to treat its own instructions and tools as a fluid codebase, enabling it to write new capabilities, fix bugs in its logic, and manage its long-term memory structures. By using this within Openclaw Skills, you enable a continuous improvement loop where the agent evolves based on every interaction.
Self-Evolve: Autonomous Agent Evolution Protocol Use Cases
- Proactively identifying and fixing ambiguous prompts in the system configuration.
- Developing and publishing new skills to the registry when gaps in capability are detected.
- Automatically updating long-term memory structures to improve context retrieval efficiency.
- Diagnosing and patching logic bugs in existing skill workflows without developer intervention.
- Adjusting agent personality and response styles based on user satisfaction levels.
How Self-Evolve: Autonomous Agent Evolution Protocol Works
- The agent monitors its own performance during heartbeat intervals or idle cycles to identify weaknesses.
- It performs a root cause analysis by reviewing interaction logs and memory files to find areas for improvement.
- The agent autonomously modifies configuration files, refines prompts, or generates entirely new skills in the workspace.
- Changes are applied immediately to the agent's runtime environment without requiring a manual restart or user confirmation.
- Every modification is logged in a standardized evolution diary within the memory directory to allow for transparency and potential rollbacks.
Self-Evolve: Autonomous Agent Evolution Protocol Setup
To enable autonomous evolution for your agent, install the skill into your local workspace environment:
npx clawhub install self-evolve
Ensure that the agent has sufficient filesystem permissions to modify the Openclaw Skills configuration files located in ~/.openclaw/ and the project root.
Self-Evolve: Autonomous Agent Evolution Protocol Data Schema & Taxonomy
The skill manages a specific taxonomy of files to facilitate self-improvement:
| File | Purpose | Evolution Impact |
|---|---|---|
| SOUL.md | Personality & Identity | Adjusts tone, verbosity, and persona |
| AGENTS.md | Behavioral Rules | Optimizes execution logic and task handling |
| MEMORY.md | Knowledge Base | Refines how long-term data is structured |
| openclaw.json | System Config | Updates tools, model parameters, and hooks |
| skills/ | Custom Skills | Generates new SKILL.md and skill.json files |
Self-Evolve: Autonomous Agent Evolution Protocol Advanced Features
- Automated skill publishing using the npx clawhub publish command for community contribution.
- Self-diagnostic hooks that trigger when repetitive errors are detected in the terminal output.
- Full bash and curl integration for fetching external documentation to fill internal knowledge gaps.
- Dynamic heartbeat task modification to schedule its own maintenance and optimization routines.
- Multi-agent orchestration tuning by modifying the dispatching rules in the core configuration files.
SKILL.md
Loading
Related Openclaw Skills
Respond-First Multi-Agent Dispatcher
A sophisticated multi-agent coordinator that delegates complex tasks to a pool of five specialized sub-agents using round-robin scheduling.
Naruto Multi-Agent Dispatcher
A Naruto-themed multi-agent system that ranks and dispatches missions to specialized sub-agents for efficient task execution.
Unified Skill Finder
A comprehensive search and installation tool for discovering agent capabilities across multiple major skill registries.
Mac Contacts Manager
A specialized CLI tool for reading, creating, and managing macOS Contacts via the native CNContactStore API.
SnapOG
SnapOG is a high-performance API for generating professional social media images, OG cards, and marketing visuals from customizable templates.
GoldenSeed
GoldenSeed provides deterministic entropy streams from fixed seeds, ensuring perfectly reproducible randomness for testing and procedural generation.
Featured*
