AgentSkillsCN

Skill Evolution Manager

一款核心工具，专门用于根据用户反馈和对话内容，在对话结束时对现有技能进行总结、优化与迭代。它不断汲取对话中的“精华”（如成功的解决方案、失败的经验教训、特定的代码规范），从而持续完善技能库。

SKILL.md
--- frontmatter
name: Skill Evolution Manager
description: A core tool specifically designed to summarize, optimize and iterate existing Skills based on user feedback and conversation content at the end of a conversation. It continuously evolves the Skills library by drawing on the "best" of the conversation (such as successful solutions, failed lessons, specific code specifications).
license: MIT

Skill Evolution Manager

This is the "evolutionary center" of the entire AI skill system. It is not only responsible for optimizing a single Skill, but also responsible for reviewing and accumulating experience across Skills.

Core responsibilities

  1. •Session Review: At the end of the conversation, analyze the performance of all called Skills.
  2. •Experience Extraction: Convert unstructured user feedback into structured JSON data (evolution.json).
  3. •Smart Stitching: Automatically write accumulated experience to SKILL.md to ensure persistence and is not overwritten by version updates.

Usage scenarios

触发：

  • •/evolve
  • •"Review the conversation just now"
  • •"I don't think the tool just now is easy to use. Let's record it."
  • •"Save this experience to Skill"

工作流（The Evolution Workflow）

1. Review & Extract

When the user triggers a review, the Agent must execute:

  1. •Scan context: Find the points where the user is dissatisfied (error, wrong style, wrong parameters) or the points where they are satisfied (specific Prompt works well).
  2. •Location Skill: Determine which Skill needs to evolve (such as yt-dlp or baoyu-comic).
  3. •Generate JSON: Build the following JSON structure in memory:
json
{ 
"preferences": ["User wants download by default mute"], 
"fixes": ["ffmpeg path needs to be escaped under Windows"], 
"custom_prompts": "Always print the estimated time before execution" 
}

2. Experience Persistence

The Agent calls scripts/merge_evolution.py to write the above JSON increment to the evolution.json file of the target Skill.

  • •Command: python scripts/merge_evolution.py <skill_path> <json_string>

3. Document Stitch

The Agent calls scripts/smart_stitch.py to convert the contents of evolution.json to Markdown and append it to the end of SKILL.md.

  • •Command: python scripts/smart_stitch.py ​​<skill_path>

4. Cross-version alignment (Align)

When skill-manager updates a certain Skill, the Agent should actively run smart_stitch.py to "restitch" the previously saved experience into the new version of the document.

Core script

  • •scripts/merge_evolution.py: Incremental merging tool. Responsible for reading the old JSON, re-merging the new List, and saving it.
  • •scripts/smart_stitch.py: Document generation tool. Responsible for reading JSON, generating or updating the ## User-Learned Best Practices & Constraints chapter at the end of SKILL.md.
  • •scripts/align_all.py: Full alignment tool. Traverse all Skill folders with one click and re-stitch the existing evolution.json experience back to the corresponding SKILL.md. Commonly used for experience restoration after skill-manager batch updates.

Best practices

  • •Do not modify the body of SKILL.md directly: Unless it is an obvious spelling error. All experience corrections should be made through the evolution.json channel to ensure that experience is not lost when the skill is upgraded.
  • •Multi-Skill Collaboration: If a conversation involves multiple Skills, perform the above process for each Skill in turn.