post-mortem

# Post-Mortem Skill

> **Quick Ref:** Full validation + knowledge extraction. Output: `.agents/retros/*.md` + `.agents/learnings/*.md` + Gate 4 decision.

**YOU MUST EXECUTE THIS WORKFLOW. Do not just describe it.**

Validate, learn, and feed knowledge back into the flywheel.

## Execution Steps

Given `/post-mortem [epic-id]`:

### Step 1: Identify What Was Completed

**If epic ID provided:** Use it directly.

**If no epic ID:** Find recently completed work:
```bash
bd list --status closed --since "7 days ago" 2>/dev/null | head -5
```

Or check recent git activity:
```bash
git log --oneline --since="24 hours ago" | head -10
```

### Step 1a: Pre-flight Check - Work Exists

**Verify there is work to post-mortem:**

```bash
# Count recent commits
git log --oneline --since="7 days ago" 2>/dev/null | wc -l
```

**If 0 commits found and no epic ID provided:**
```
STOP and return error:
  "No work found to post-mortem. Either:
   - Provide epic ID: /post-mortem <epic-id>
   - Or complete some work first"
```

Do NOT run post-mortem on empty work - this produces misleading "all clear" reports.

### Step 2: Run Code Validation (Inline - No Nesting)

**Read changed files and validate directly:**

```bash
# Get recent changes
git diff --name-only HEAD~5 2>/dev/null | head -20
```

Read each changed file with the Read tool, then apply the 8 vibe aspects:
1. Semantic - code matches docs?
2. Security - vulnerabilities?
3. Quality - dead code, smells?
4. Architecture - layer violations?
5. Complexity - too complex?
6. Performance - issues?
7. Slop - AI hallucinations?
8. Accessibility - issues?

Record findings with file:line citations.

### Step 2a: Early Exit on CRITICAL

**If inline validation found CRITICAL issues:**

```
STOP IMMEDIATELY. Do NOT dispatch the swarm.

Report to user:
  "CRITICAL issues found in inline validation. Fix before continuing post-mortem."

  Findings:
  - <file:line> - <critical issue>

  Action: Fix these issues, then re-run /post-mortem
```

**CR