AgentSkillsCN

learnings

从当前会话中提炼出不显而易见的学习心得，并将其自动保存至记忆中。当用户希望记录一次编码会话中学到的内容时使用此功能。

SKILL.md
--- frontmatter
name: learnings
description: "Extract non-obvious learnings from the current session and save them to auto memory. Use when the user wants to capture what was learned during a coding session."
disable-model-invocation: true

Analyze our conversation and extract non-obvious learnings, then save them to the appropriate auto memory files.

The motto is: "How can we do this better next time?"

What counts as a learning (non-obvious discoveries only):

  • •Hidden relationships between files or modules
  • •Execution paths that differ from how code appears
  • •Commands we tried and eventually got working
  • •Non-obvious configuration, env vars, or flags
  • •Debugging breakthroughs when error messages were misleading
  • •API/tool quirks and workarounds
  • •Build/test commands not in README
  • •Architectural decisions and constraints
  • •Files that must change together

What NOT to include:

  • •Obvious facts from documentation
  • •Standard language/framework behavior
  • •Things already in CLAUDE.md or MEMORY.md

Process:

  1. •Review the full conversation for non-obvious discoveries
  2. •Read the current MEMORY.md to avoid duplicating existing knowledge
  3. •Categorize learnings by topic
  4. •For each topic, either update an existing memory file or create a new one in the auto memory directory
  5. •Update MEMORY.md with a brief summary and links to topic files
  6. •Present the learnings to the user for review before saving

Keep MEMORY.md concise (under 200 lines). Use separate topic files for detailed notes.