| name | post-mortem |
| description | MUST use when user says 'post-mortem', 'postmortem', 'log this failure', 'write a post-mortem', 'what went wrong'; or auto-detect signals (failed deploys, broken builds, regressions surfaced). Captures failure with context, root cause, and learning so future sessions don't repeat it. Append-only log; no blame, only growth. |
🔥 Post-Mortem — Skill Plugin
Skill Name
Post-Mortem System
Trigger Words
"post-mortem"
"postmortem"
"log this failure"
"write a post-mortem"
"what went wrong"
Auto-Detection Triggers (Passive — Always Active)
AI watches for these signals and prompts the user:
| Signal | Phrase Examples |
|---|
| Deployment failure | "it crashed", "pod is failing", "image pull error", "rollback" |
| Test regression | "tests are broken", "was passing before", "something broke" |
| Architecture reversal | "undo this", "we need to revert", "this approach doesn't work" |
| Wasted time | "wasted hours", "dead end", "that didn't work at all" |
| Security incident | "exposed secret", "accidentally committed", "vulnerability" |
| Data loss | "data is gone", "migration failed", "backup didn't work" |
On detection, AI asks: "That didn't go as planned. Worth a post-mortem?"
User says yes → AI fills out the format from post-mortem-core.md.
User says no → move on, no log created.
Manual Trigger
User says "post-mortem" or "log this failure" → AI immediately starts the post-mortem format.
Behavior
- Detect signal (passive) or receive explicit trigger (manual)
- Ask: "Worth a post-mortem?" (skip if manual trigger — user already decided)
- If yes: fill out format from
post-mortem-core.md, ask clarifying questions as needed
- Append entry to
main/post-mortems.md
- Reference entry in future sessions when work touches the same domain
Domain Reference Behavior
When starting work in a domain that has a past post-mortem:
- Check
main/post-mortems.md for relevant entries
- Flag: "⚠️ Reminder: [lesson] — see post-mortem [date]"
Level History
- Lv.1 — Base: manual trigger + append to log
- Lv.2 — Auto-detection of failure signals + passive prompting
- Lv.3 — Domain reference: flag relevant post-mortems at session start or task start