Security audit workflow - vulnerability scan → verification
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
- Quick scan of conversation
- If nothing non-obvious → "Nothing to capture, we're good"
- If something surfaces → update
~/clawd/MEMORY.mdor today's daily note - 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.