process-meeting

name	process-meeting
description	Process a meeting transcript — extract decisions, action items, insights, and commitments. Use when the human says "process this meeting", shares a transcript from Granola/Otter/etc, or pastes meeting notes.

process-meeting

Turns a meeting transcript into structured knowledge and actionable items. Stores the source, extracts what matters, routes to the right places.

When to use

• "Process this meeting", "here's a transcript"
• Human shares output from Granola, Otter, or any meeting recorder
• After any significant conversation worth capturing

Inputs

Input	Required	Description
transcript	yes	Meeting text — pasted, file path, or URL
context	no	Meeting name, participants, project it relates to

Process

STEP 1 — Store source

Save the raw transcript to superpaper/sources/meetings/ with frontmatter:

type: source
source: meeting
participants: []
created: YYYY-MM-DD

STEP 2 — Extract

Read the full transcript. Retrieve knowledge neighborhood for mentioned concepts, people, and projects. Extract:

1. Decisions — what was decided and by whom
2. Action items — who committed to what, with deadlines if mentioned
3. Insights — ideas, frameworks, connections worth remembering
4. Open questions — unresolved topics that need follow-up
5. Commitments — promises made (by human or others)

STEP 3 — Route (atomic-first)

Extract aggressively — every distinct concept, quote, decision, and finding deserves its own atomic note. A quote by a participant → its own note linked to that person's people/ entry. A framework or mental model mentioned → its own concept note. The more atoms, the richer the connection network.

• Decisions → append to [[Decisions log]] with date, participants, reasoning. If the decision reasoning is substantial, also create an atomic type: decision note.
• Insights worth keeping → create atomic notes in superpaper/concepts/. One concept per note. Link to source transcript (![[transcript#^passage]]) and related existing notes. Preserve attribution — who said it, in what context.
• People context → create or update people/ notes for participants. Add new context, relationship info, or notable quotes (as their own linked notes).
• Learnings → append to [[Learning log]]
• Action items → add to the task board (apps/My tasks.md) under Todo
• Project-relevant items → update the relevant project note in superpaper/projects/
• If the meeting produced 3+ atomic notes, create a lightweight hub note that embeds them with connective prose — this becomes the "processed meeting summary."

STEP 4 — Agent log summary

Append to today's agent log (inbox/log/YYYY-MM-DD):

### Meeting: [name] — HH:MM
- **With:** [participants]
- **Decisions:** [1-line each]
- **Actions:** N items added to board
- **Insights:** [[note1]], [[note2]]
- **Follow up:** [open questions]

Never write to the human's daily note (daily/YYYY-MM-DD) — it's a pure date anchor.

Outputs

• Source transcript in superpaper/sources/meetings/
• Updated living documents (Decisions log, Learning log)
• 0+ new atomic notes in entity folders
• Action items on task board
• Agent log meeting summary (inbox/log/YYYY-MM-DD)

Decision authority

• YOU DECIDE: what to extract, severity/priority of action items, routing
• ESCALATE TO HUMAN: anything involving commitments to others, sensitive content, ambiguous decisions

All notes created by this skill must include created-by: ai. The meeting transcript itself is created-by: human — never modify its body, only update frontmatter properties. Source notes, atomic notes, log entries, and board cards are created-by: ai.