| name | epic-postmortem |
| description | Create or update epic retrospective documents. Use when user mentions "postmortem", "post-mortem", "post mortem", "retrospective", or "retro" for a completed epic or feature. Distinct from workflow-postmortem which tracks execution issues. |
Epic Postmortem Skill
Purpose: Document learnings from completed epics/features for team improvement
Output: docs/epics/backlog/{project-code}-{epic-code}-post-mortem.md in metasaver-marketplace repo
Modes
| Mode | When to Use | Action |
|---|
create | New post-mortem for epic | Generate from template |
update | Add learnings to existing | Append to relevant sections |
Workflow
Mode: create
- Gather info: Ask user for project code, epic code, epic description
- Validate: Check file does not already exist at target path
- Generate: Use template from
templates/post-mortem-template.md
- Interview: Prompt user for each section (what went well, wrong, learnings)
- Write: Save to
docs/epics/backlog/{project}-{epic}-post-mortem.md
- Confirm: Report file path to user
Mode: update
- Locate: Find existing post-mortem by project-epic code
- Read: Load current content
- Interview: Ask user what new learnings to add
- Append: Add entries to appropriate sections with timestamp
- Save: Write updated file
- Confirm: Report changes to user
File Naming Convention
Pattern: {project-code}-{epic-code}-post-mortem.md
| Component | Format | Example |
|---|
| project-code | 2-4 chars | msc |
| epic-code | 2-4 chars | lct |
Examples:
msc-lct-post-mortem.md (MetaSaver Com - Listing Contacts)chb-air-post-mortem.md (Chatbot - AI Routing)
Output Location
Repository: metasaver-marketplace (centralized)
Path: docs/epics/backlog/
Template Sections
See templates/post-mortem-template.md for full structure:
| Section | Purpose |
|---|
| Epic Info | Project, code, description, dates |
| What Went Well | Successes to repeat |
| What Went Wrong | Failures to avoid |
| Learnings | Actionable improvements for future |
| Action Items | Specific follow-ups with owners |
Examples
Create new post-mortem:
User: "Create a postmortem for the MSC listing contacts epic"
Agent:
1. Asks: "What is the epic code?" -> "lct"
2. Asks: "Brief description?" -> "CRUD for listing contacts"
3. Asks: "What went well?" -> user provides bullet points
4. Asks: "What went wrong?" -> user provides bullet points
5. Asks: "Key learnings?" -> user provides bullet points
6. Asks: "Action items?" -> user provides with owners
7. Writes: docs/epics/backlog/msc-lct-post-mortem.md
Update existing post-mortem:
User: "Add learnings to the MSC-LCT post-mortem"
Agent:
1. Reads: docs/epics/backlog/msc-lct-post-mortem.md
2. Asks: "What new learnings to add?"
3. Appends: New entries with timestamp
4. Saves: Updated file
Distinction from workflow-postmortem
| Aspect | epic-postmortem | workflow-postmortem |
|---|
| Scope | Project/epic retrospective | Single workflow execution |
| Content | Team learnings, process issues | Agent mistakes, skipped steps |
| Location | docs/epics/backlog/ | docs/projects/{folder}/ |
| Trigger | User request after epic done | Automatic during /build, /ms |
| Audience | Team/stakeholders | Developer debugging workflows |
