Post-Mortem

Install in one command

mkdir -p ~/.claude/skills/post-mortem && curl -fsSL https://eliteaiadvantage.com/skills/post-mortem/SKILL.md -o ~/.claude/skills/post-mortem/SKILL.md

Post-Mortem turns a freshly-resolved bug or incident into a structured memory file inside your project's Claude Code memory directory. The next session you start, Claude already knows the lesson, including the specific trigger conditions for when to recall it.

The output is not a doc you read once. It's a memory artifact with a "When to remember this" section at the bottom that tells future Claude sessions exactly when this lesson applies. Without it, painful bug lessons evaporate within a week. With it, the next time a similar pattern shows up, you skip the rediscovery.

• →Captures the lesson while it's still vivid, not three weeks later when half the details are gone.
• →The "When to remember this" section turns the file into actual retrieval-grade memory.
• →Updates the project MEMORY.md index automatically so the file is discoverable.
• →Severity-graded (minor / moderate / painful / disaster) so write-up depth matches incident size.
• →Tone is factual, not apologetic, no "we should have caught this" hand-wringing.

• •Just resolved a bug that took more than 30 minutes to track down.
• •Identified a design pattern that allowed the bug, capture before it fades.
• •A production incident on a queue, cron, or deploy.
• •A migration that went wrong (or almost did) and is worth remembering.

Without this skill, every Claude session starts cold, and bug patterns that took you hours to debug last month are unknown to today's session. Post-Mortem writes a structured feedback file with explicit retrieval triggers, so when you (or future Claude) brushes up against the same pattern, the memory surfaces. Claude stops re-discovering bugs you already paid for.