AgentSkillsCN

checkout

自动结束会话——无需任何提示,只需捕获并展示结果。

SKILL.md
--- frontmatter
name: checkout
description: Automated session closing - no prompts, just capture and display.
license: MIT

ELF Checkout Command

Automated session closing - no prompts, just capture and display.

What It Does

The /checkout command:

  • Auto-detects session activity (commits, files, domains)
  • Counts heuristics recorded during the session
  • Displays a summary
  • No questions asked

Usage

bash
/checkout

Execution

bash
python ~/.claude/emergent-learning/src/query/checkout.py

Output Example

code
┌────────────────────────────────────┐
│    Emergent Learning Framework     │
├────────────────────────────────────┤
│                                    │
│      Session Complete              │
│      Auto-capturing learnings...   │
│                                    │
└────────────────────────────────────┘

[*] Session Summary (auto-detected)
   Domains: infrastructure, backend
   Commits: 3
   Files modified: 7
     - src/query/checkin.py
     - src/query/checkout.py
     ... and 5 more
   Heuristics recorded: 2

[OK] Checkout complete.

What Gets Captured

  • Domains: Auto-detected from file paths (dashboard, frontend, backend, infrastructure)
  • Commits: Count of recent commits in the repo
  • Files: List of recently modified files
  • Heuristics: Count from database (last 4 hours)

Philosophy

Checkout should be frictionless. Record learnings during the session using:

  • [LEARNED:domain] pattern markers in subagent outputs
  • record-heuristic.py script when you discover something
  • Failure analysis files when things break

Checkout just summarizes what happened - it doesn't interrogate you at the end.

Complementing /checkin

CommandPurpose
/checkinLoad context at session start
/checkoutShow session summary at end