ai-self-evolution
记录经验、错误与修正，持续改进。触发场景：命令失败 | 操作出错 | 用户纠正（不对、实际上、你错了） | 功能请求（能不能、我希望、有没有办法） | API或工具失败 | 知识过时 | 发现更优做法 | 重复模式 | 非显而易见的问题。执行重大任务前先回顾历史经验。会话开始时回顾，会话结束时总结。
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/andylue/ai-self-evolutionWhat This Skill Does
The ai-self-evolution skill serves as the cognitive memory and improvement engine for the OpenClaw AI agent. It implements a formal feedback loop by monitoring, logging, and processing session interactions—such as errors, user corrections, and feature requests—into structured markdown files located in the .learnings/ directory. By performing automated reviews at the start of tasks and summarizing experiences at the end, the agent transitions from a stateless tool to a persistent assistant that learns from its environment. It effectively automates the 'simplify-and-harden' methodology, allowing the agent to refine its workflows, adopt best practices, and consolidate project-specific knowledge over time.
Installation
To integrate this skill into your environment, execute the following command in your terminal:
clawhub install openclaw/skills/skills/andylue/ai-self-evolution
Ensure that your environment allows for file-system operations, as this skill relies on the creation and maintenance of logs within the .learnings/ directory.
Use Cases
- Continuous Refinement: When a user points out an error or provides a better way to execute a task, the agent logs this as a
correctionorbest_practiceto prevent future recurrence. - Error Tracking: Automatically captures stack traces, non-zero exit codes, and API failures, providing a historical log for debugging persistent environment issues.
- Feature Lifecycle Management: Captures user feature requests directly within the project workflow, ensuring that ideas are not lost between sessions.
- Knowledge Retention: Maintains internal documentation (
AGENTS.md,CLAUDE.md) by promoting high-value learnings to shared knowledge repositories.
Example Prompts
- "Wait, that approach to the API request is outdated; you should use the new auth header instead."
- "Why did that deployment script fail? Can you check the logs and record what happened?"
- "I'd love for you to remember to always format the output as JSON for this project from now on."
Tips & Limitations
- Tip: Keep your
Pattern-Keynaming consistent to allow the automatic merger to consolidate repetitive errors effectively. - Tip: Review the
.learnings/directory occasionally to ensure that the agent isn't cluttering the workspace with redundant logs. - Limitation: The skill is reactive; it requires the agent to be active to 'learn.' If a system crash occurs outside the agent's control, manual entry into
ERRORS.mdmay be required. - Maintenance: Ensure your environment has sufficient disk space for long-term logs; the system handles cleanup, but large-scale projects can generate significant historical data.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-andylue-ai-self-evolution": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-write, file-read