daily-evolution

What This Skill Does

The daily-evolution skill, also known as the "Hidden Dragon Project," serves as the core intelligence engine for an AI Agent to achieve continuous self-improvement. By automating a comprehensive four-phase review process, the agent acts as an autonomous collaborator that systematically refines its behavior based on past interactions. The engine operates through a structured cycle: it begins by auditing the previous 24 hours of conversation to extract high-value insights, follows up with a deep-dive scan of tool health and dependencies, proceeds to solidify successful strategies into core configuration files like SOUL.md and TOOLS.md, and concludes by generating a detailed execution report. This creates a feedback loop where the agent not only executes tasks but learns from the operational patterns, optimizes its own toolchain, and keeps its persona aligned with the evolving needs of the user.

Installation

To integrate this skill into your environment, use the OpenClaw command-line interface. Run the following command in your terminal:

clawhub install openclaw/skills/skills/wilsonliu95/daily-evolution

Ensure that your workspace directory has write permissions enabled, as the skill will automatically generate reports in the workspace/reports/ directory and perform maintenance updates on system metadata files.

Use Cases

• Long-term Project Management: Perfect for agents that manage recurring tasks where workflow bottlenecks are common. The agent will identify and 'self-patch' its processes to prevent future errors.
• Persona Fine-tuning: Ideal for users who want their agent to gradually 'grow' into their personal communication style, remembering specific preferences and habits without needing constant manual reminders.
• Autonomous Tool Maintenance: Useful for development environments where the agent manages multiple MCP services and API integrations, automatically flagging dead keys or unused tools.

Example Prompts

1. "执行潜龙计划" (Execute the Hidden Dragon Project to start a full evolution cycle)
2. "执行每日进化" (Execute daily evolution to update my persona based on today's collaboration)
3. "查看进化报告" (Display the latest evolution report to see what the agent learned today)

Tips & Limitations

• Scheduling: While it works best with a 00:00 UTC+8 cron trigger, avoid scheduling it during high-intensity usage periods to ensure the agent has clean access to its own files.
• Safety: The agent is restricted from modifying project code or user data; it exclusively edits its own 'soul' (persona and tool settings). Always check the diff logs generated during Phase 3 to audit what changes the agent made to its configuration. If the agent makes a change that doesn't fit your needs, simply revert the specific file modification.