A meta-skill designed to architect, deploy, and continuously evolve autonomous AI agents following a standardized self-improvement model.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install evo-skill-creator
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 evo-skill-creator 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 Evo Skill Creator?
Evo Skill Creator is a sophisticated meta-intelligent agent designed specifically for the creation and management of other AI agents. It operates as an evolutionary engine, ensuring that every agent it produces is equipped with the capability for self-learning, trend scanning, and performance review. By utilizing this skill, developers can maintain a consistent architecture across their entire suite of Openclaw Skills, allowing for a unified ecosystem of agents that grow more powerful over time.
The skill is built on a specialized evolution model that bridges the gap between static task execution and dynamic domain expertise. It doesn't just create a persona; it builds a full operational framework including memory structures, learning plans, and execution logs, ensuring that the resulting agents are professional, technical, and highly actionable from the moment of initialization.
Evo Skill Creator Use Cases
- Rapidly architecting new domain-specific AI agents with standardized memory structures.
- Updating and upgrading existing agent frameworks to support self-evolution protocols.
- Managing a centralized registry of multiple AI skills within a single workspace.
- Automating the acquisition of new knowledge and trend scanning for AI personas.
- Batch-reviewing and refining the quality of an entire agent ecosystem.
How Evo Skill Creator Works
- Initialization: The agent reads the scene index and private memory to understand the current context and available sub-agents.
- Command Processing: It accepts natural language or command-line triggers such as go, learn, scan, and review.
- Alignment: Before executing complex tasks, it follows the 3W Principle (Why, What, How) to ensure the proposed solution aligns with the user's goals.
- Generation: It generates a complete Skill package, including the SKILL.md, .gitignore, and necessary directory structures for memory and output.
- Registration: The new agent is recorded in a private registry to facilitate future batch updates and status monitoring.
- Evolution: Through the review command, it can propagate its own structural improvements to all previously created sub-agents.
Evo Skill Creator Setup
To install and configure this skill within your Openclaw Skills environment, follow these steps:
- Copy the skill directory into your local skills folder.
- Configure your permissions in
.claude/settings.local.jsonusing the provided template. - Initialize the agent and check its status:
/evo-skill-creator status
- To create your first self-evolving agent, use the go command:
/evo-skill-creator go "create a python expert agent"
Evo Skill Creator Data Schema & Taxonomy
The skill follows a strict directory structure to separate public knowledge from private execution data within Openclaw Skills:
| Path | Purpose |
|---|---|
memory/learning-plan.md |
Tracks the agent's long-term educational objectives. |
memory/knowledge/ |
A categorized repository for domain-specific knowledge (AI models, frameworks, etc.). |
memory/private/backlog.md |
Stores pending tasks and priority levels for the agent. |
memory/private/agents-registry.md |
A markdown table tracking all sub-agents, their versions, and paths. |
output/execution.log |
A chronological record of every command executed by the agent. |
output/report/ |
Contains timestamped markdown reports for every major action (go, learn, scan, plan). |
Evo Skill Creator Advanced Features
- Gene Replication: A unique mechanism to distribute the creator's latest evolutionary upgrades to all child agents.
- Autonomous Trend Scanning: Automatically identifies and incorporates new AI methodologies into the learning plan.
- Structured Self-Reflection: Built-in self-audit protocols to check for goal alignment and output quality before delivery.
- Multi-Agent Registry: Centralized management of complex agent ecosystems including versioning and status tracking.
- Flexible Data Pathing: Support for relative and absolute pathing to ensure portability across different development environments.
SKILL.md
Loading
Related Openclaw Skills
Cool Skill
A creative suite designed to inject style, personality, and flair into your coding projects and documentation.
CLAUDE.md Auto-Discovery Skill
Automatically discover and inject project-specific instructions from CLAUDE.md files into your AI agent context.
GitHub Operations Skill
A fully autonomous GitHub management skill that enables AI agents to create repositories, push code, and manage releases without user intervention.
Test Skill
A specialized utility designed to benchmark and validate the publishing workflow and structural integrity of Openclaw Skills.
Feishu Skills Installer
A comprehensive installer for deploying the full suite of Feishu (Lark) integration capabilities to your AI agent.
shadcn-ui Expert
A specialist implementation skill for building production-grade user interfaces using shadcn/ui, Tailwind CSS, and robust form validation patterns.
Featured*
