transcript triage

Transcript Triage

Parse AI conversation transcripts → numbered topics → groom into positions.

Flow

1. Trigger: User provides transcript (MD format)
2. Parse: AI extracts topics, decisions, action items
3. Triage: Classify by urgency/importance:
   • NOW - Current epic (immediate)
   • LATER - Future epics (planned)
   • BACKLOG - Ideas (someday/maybe)
   • DECISIONS - What was agreed (log)
4. Groom: User positions topics in correct places (ROADMAP, epic-notes, memory)

Triggers

• "triage transcript"
• "parse conversation"
• "organize chat"
• "extract action items from [file/paste]"

Output Format

## Triage Results

### NOW (v1.1.0 Contract Diagram)
1. [Topic] - [Brief description]
2. [Topic] - [Brief description]

### LATER (Future Epics)
3. [Topic] - [Brief description]
4. [Topic] - [Brief description]

### BACKLOG (Ideas)
5. [Topic] - [Brief description]

### DECISIONS (Log to Memory)
- [Decision] - [Context]
- [Decision] - [Context]

Integration Points

• Backstage: Auto-add NOW items to current epic-notes/
• ROADMAP: Suggest LATER items as new epics
• Memory: Log DECISIONS to memory/YYYY-MM-DD.md

Notes

Why this skill?

• External AI chats (Claude, ChatGPT, etc.) generate valuable insights
• Transcripts are hard to parse manually
• Need systematic way to extract → classify → act

Similar skills:

• Backstage (epic management, checks)
• Roadmap (epic planning, grooming)
• Memory (decision logging)

Unique value:

• Bridges external conversations → internal workflow
• Automates triage (urgency + importance)
• Preserves context (decisions logged, not lost)