A recursive self-evolution engine that allows AI agents to autonomously analyze their runtime history, fix bugs, and optimize their own source code.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install evomap-evolver
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 evomap-evolver 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 Capability Evolver?
The Evolver is a sophisticated meta-skill designed for the Openclaw Skills ecosystem, acting as a recursive self-improvement engine. It empowers agents to transcend static capabilities by inspecting their own runtime history, identifying failures or inefficiencies, and autonomously writing new code or updating memory to improve performance. This system ensures that evolution is a continuous process, allowing agents to adapt to new challenges without manual developer intervention.
Built on the Gene-Evolution Protocol (GEP), this skill provides a standardized framework for auditable evolution. It treats code and logic updates as reusable assets, ensuring that successful reasoning patterns are preserved while failures are systematically repaired. By integrating this into your Openclaw Skills library, you enable a lifecycle of constant optimization and autonomous resilience.
Capability Evolver Use Cases
- Autonomous bug detection and self-repair for failing agent workflows.
- Performance optimization and code refactoring during system idle time.
- Self-healing of dependency issues and missing library requirements.
- Continuous improvement of agent memory and reasoning logic based on task history.
How Capability Evolver Works
- The system scans the workspace memory and narrative logs to detect errors, crashes, or performance patterns.
- It selects an evolution strategy—such as repair, innovate, or harden—based on the current stability of the agent.
- Using LLM-driven analysis, it generates code patches or memory updates that address the identified issues.
- The changes are applied to the source code or assets, using Git-based version control to ensure safety and provide rollback capabilities if the evolution fails.
Capability Evolver Setup
To get started with this evolution engine for your Openclaw Skills, register your node and configure your environment:
- Register your node identity via the EvoMap network and obtain your ID.
- Set the required environment variable:
export A2A_NODE_ID=node_xxxxxxxxxxxx
- Run the evolution cycle in automated mode:
node index.js
- For safety, it is recommended to run this in a directory managed by Git to support the automatic rollback features.
Capability Evolver Data Schema & Taxonomy
The skill organizes its evolution data through a structured repository of assets and memory files:
| Path | Format | Description |
|---|---|---|
assets/gep/genes.json |
JSON | Definitions of reusable logic components known as Genes. |
assets/gep/capsules.json |
JSON | Success capsules that store reasoning to prevent repeating errors. |
assets/gep/events.jsonl |
JSONL | An append-only log of evolution events and parent-child relationships. |
memory/ |
Directory | Contains the knowledge graph, narrative logs, and reflection history. |
src/** |
Source | The evolved code base, modified only during the solidification phase. |
Capability Evolver Advanced Features
- Mad Dog Mode: Enables a continuous loop for infinite autonomous self-improvement cycles.
- Review Mode: Allows human-in-the-loop verification before any code changes are solidified.
- GEP Protocol: A constrained evolution prompt system that ensures changes are auditable and structured.
- Automatic Rollbacks: Supports hard or stashed git resets to recover from failed evolution attempts.
- Cross-Skill Integration: Automatically detects and utilizes other Openclaw Skills like feishu-card for advanced reporting.
SKILL.md
Loading
Related Openclaw Skills
CNOOC Brand & Industry Intelligence
A specialized knowledge skill providing structured information and professional brand intelligence on CNOOC for AI agents.
CNPC/PetroChina Information Assistant
A comprehensive digital assistant providing real-time data on CNPC/PetroChina products, financial performance, and official resources.
Coachella Music Festival Skill
A comprehensive encyclopedia skill providing real-time information, history, and key statistics for the Coachella Valley Music and Arts Festival.
Coconut 椰子 Encyclopedia Guide
A comprehensive digital encyclopedia for all things coconut, covering history, nutrition, and global sourcing.
Awesome Closet Stylist
A high-autonomy AI agent skill that transforms your digital wardrobe into a personal stylist through natural language interaction and intelligent outfit planning.
Hyperliquid Names (HLN) API
An AI agent skill designed to interact with the Hyperliquid Names API for domain resolution, profile lookups, and HyperEVM integration.
Featured*
