terry-li-hm

retro

0
0
# Install this skill:
npx skills add terry-li-hm/skills --skill "retro"

Install specific skill from multi-skill repository

# Description

End-of-session retrospective. Scan for non-obvious learnings that might have slipped through. Use when user says "retro", "wrap up", "what did we learn", or at end of meaty sessions.

# SKILL.md


name: retro
description: End-of-session retrospective. Scan for non-obvious learnings that might have slipped through. Use when user says "retro", "wrap up", "what did we learn", or at end of meaty sessions.


Retro

Quick cleanup sweep at end of session — catch what "Compound As You Go" missed.

Triggers

  • "retro", "wrap up", "let's wrap"
  • "what did we learn"
  • End of long/meaty session

When to Skip

  • Session was trivial (quick questions, nothing complex)
  • Already captured learnings during the session
  • User seems done, don't force it

What to Look For

Scan for non-obvious stuff only:

Type Signal
Patterns Same issue came up 3 times — that's a pattern
Implicit preferences Terry kept choosing X over Y — preference?
Hidden friction Something took 4 attempts — why?
Surprising wins That worked way better than expected — why?
Mistakes I missed Wrong assumption I didn't notice until now

Workflow

  1. Quick scan of conversation
  2. If nothing non-obvious → "Nothing to capture, we're good"
  3. If something surfaces → update ~/clawd/MEMORY.md or today's daily note
  4. Done. No ceremony.

Output

If something found:

**Retro:**
- [Learning] → added to MEMORY.md

If nothing:

Nothing non-obvious this session. ✓

Notes

  • This is a sweep, not a ritual
  • Obvious corrections should've been captured on-the-fly
  • Focus on patterns and implicit signals

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.