# summarize-meetings

Adopt this skill: https://proskills.md/skills/other/2389-research-summarize-meetings-skills

Page: https://proskills.md/skills/other/2389-research-summarize-meetings-skills
This document is the public skill bundle for that page. It contains SKILL.md and every helper file the catalog carries. No login is required.

## File: skills/SKILL.md
size: 8028
upstream: https://github.com/2389-research/summarize-meetings/blob/main/skills/SKILL.md

````text
---
name: summarize-meetings
description: Use when processing, summarizing, or extracting insights from meeting notes in the Obsidian vault. Triggers on "summarize meetings", "process meetings", "meeting summaries", "extract from meetings", "meeting insights".
---

# Summarize Meetings

Process meeting transcripts from `Meetings/transcripts/` — extract people, action items, project ideas, blog ideas, knowledge graph connections, concepts, and general ideas. Populate the vault knowledge graph and generate rich summary files.

## Campaign Workflow

Processing works in **monthly batches**, most-recent month first. Each month follows this cycle:

```dot
digraph campaign {
  rankdir=TB;
  "Scan month (ls -lS)" -> "Triage: check small/suspicious files";
  "Triage: check small/suspicious files" -> "Identify skips vs substantive";
  "Identify skips vs substantive" -> "Dispatch wave of agents (10-15)";
  "Dispatch wave of agents (10-15)" -> "Wait for returns";
  "Wait for returns" -> "Dispatch remaining agents";
  "Dispatch remaining agents" -> "All agents returned";
  "All agents returned" -> "Compile monthly report table";
  "Compile monthly report table" -> "Post social media update";
  "Post social media update" -> "Next month or stop";
}
```

### Step 1: Scan the Month

```bash
ls -lS "Meetings/transcripts/" | grep "YYYY-MM"
```

List all transcripts for the target month, sorted by size. This gives you:
- Total count of transcripts
- File sizes (critical for triage and agent instructions)
- Filenames for slug generation

### Step 2: Check for Existing Summaries

```bash
ls "Meetings/" | grep "YYYY-MM.*summary"
```

Skip any transcript that already has a corresponding summary file.

### Step 3: Triage

Before dispatching agents, manually check files that are suspicious:

| Signal | Action |
|--------|--------|
| **< 2K** | Read the file — likely empty stub or sparse notes |
| **2K-6K** | Skim the file — may be scheduling fragment, garbled recording, or logistics-only |
| **Filename says "untitled"** | Almost always empty — read to confirm |
| **Known non-meeting patterns** | Medical appointments, kid brainstorming, screen-sharing setup, recording process discussions |

**Skip criteria (with log entry):**
- Empty stubs (template only, no content)
- < 20 words of real content after frontmatter
- Garbled/corrupted recordings
- Scheduling fragments (just logistics, no substance)
- Medical/personal appointments
- Kid content (school brainstorming, etc.)
- Recording process discussions (just setting up Granola)
- Screen-sharing setup conversations

**Sparse but substantive notes** (like handwritten bullet points from a real meeting) should still be processed — even 1.2K of real fundraising notes is worth a summary.

### Step 4: Dispatch Parallel Agents

Spawn one Task agent per transcript using `subagent_type: general-purpose` with `mode: bypassPermissions` and `run_in_background: true`.

**Wave sizing:** Dispatch 10-15 agents per wave. Wait for returns, then dispatch remaining agents. This prevents overwhelming the system while maintaining parallelism.

**Large file instructions:** For files > 50K, include "read in chunks using offset/limit" in the agent prompt. For files > 100K, explicitly say "this is a large file — read in chunks of 500-1000 lines using offset/limit parameters."

**Agent prompt template:**

```
You are a meeting summarizer for Doctor Biz's Obsidian vault at `/Users/harper/Public/Notes/Harper Notes/`.

Read the transcript at `[PATH]` ([SIZE] — [read in chunks if large]).

Then:
1. Extract: People, Action Items, Project Ideas, Blog Ideas, Knowledge Graph, Concepts, Ideas
2. Write summary to `Meetings/YYYY-MM-DD-slugified-name-summary.md`
3. Create/update People notes in `People/Firstname Lastname.md` (Title Case) — check if they exist first
4. Create Concept notes in `Concepts/` for genuinely reusable insights

Use the full summary template with YAML frontmatter (title, date, tags, type, source, status, related).
Write a strong 2-4 paragraph narrative meeting summary.
Use `[[wiki-links]]` throughout.
Never fabricate info not in the transcript.
Don't create People stubs for first-name-only references.
```

### Step 5: Compile Monthly Report

After all agents return, present a summary table:

```
| Meeting | Date | People | Actions | Concepts | Status |
|---------|------|--------|---------|----------|--------|
| harper-john | Jan 31 | 1 | 5 | 3 | Done |
| untitled | Jan 15 | - | - | - | Skipped (empty) |
```

Include: total processed, total skipped (with reasons), and notable highlights (interesting concepts, key people).

### Step 6: Social Media Update

Post a completion update via `mcp__socialmedia__create_post` with:
- Month completed
- Count of summaries processed
- Notable highlights (key meetings, interesting concepts)
- Tags: `["meeting-summaries", "month-year", "vault-curation"]`

---

## Transcript Format: Granola

Most transcripts come from Granola and have this structure:

```yaml
---
doc_id: uuid
source: granola
created_at: ISO timestamp
title: Meeting Title
participants:
- Person Name
duration_seconds: N
labels: []
creator: '[[Harper Reed]]'
attendees:
- '[[Person Name]]'
summary_text: |-
  ### Section Header
  - Bullet points of AI-generated summary
generator: muesli 1.0
---

# Meeting Title

_Date: YYYY-MM-DD_

## Participants
- **Name**, Role, Company, (email)

## Notes
(Granola-captured notes — may be sparse or detailed)

---

## Transcript

Speaker 1: ...
Speaker 2: ...
```

**Key fields to use:**
- `attendees` — canonical list of who was there
- `summary_text` — Granola's AI summary (useful cross-reference but don't rely solely on it)
- `participants` — sometimes has role/company info
- The transcript body has the real content

**Common transcript quirks:**
- Speaker attribution may be missing or wrong
- Names may be misspelled or use nicknames
- Some transcripts lack `summary_text` entirely
- Very long transcripts (100K+) may have repetitive content from multi-hour sessions

---

## Determining "Substantive"

A note is **not substantive** if:
- Body content (after frontmatter) is only template skeleton with no filled-in content
- Body has fewer than 20 words of actual content (excluding headings, `---`, empty bullets)
- All discussion/notes sections are blank
- Content is garbled/corrupted (random characters, no coherent sentences)
- Content is purely logistics (scheduling, screen sharing setup)

Skip these with a log entry. Don't waste a summary on nothing.

---

## Extraction Categories

For each substantive meeting note, extract:

### 1. People
- Full names mentioned in content or attendees
- Role/company/context if discernible
- Relationship to Doctor Biz if clear

### 2. Action Items
- Explicit tasks (`- [ ]`, "need to", "should", "will", "action:")
- Owner if identifiable
- Due dates if mentioned
- Preserve existing Obsidian Tasks syntax items

### 3. Project Ideas
- Products, tools, apps, or ventures discussed
- Both existing projects and new ideas
- Brief description of what was discussed about each

### 4. Blog Ideas
- Topics worth writing about publicly
- Interesting takes or contrarian views expressed
- Experiences worth sharing

### 5. Knowledge Graph Ideas
- Connections between entities (people <-> projects, concepts <-> people)
- Relationships worth capturing as wiki-links
- Patterns across meetings (if processing a batch)

### 6. Concepts
- Mental models, frameworks, or reusable ideas
- Strategic insights or principles discussed
- Technical concepts worth an atomic note
- **Must be genuinely reusable** — not just a topic that was discussed
- **Check vault first** — enrich existing concept notes rather than creating duplicates

### 7. Ideas
- General ideas that don't fit the above categories
- Shower thoughts, brainstorms, what-ifs
- Anything interesting that was floated

---

## Vault Graph Updates

### People Notes

For each person extracted:

1. **Check** if `People/Firstname Lastname.md` exists (search c
````

## File: README.md
upstream: https://github.com/2389-research/summarize-meetings/blob/main/README.md

````text
# Summarize Meetings

Meeting transcripts pile up. The names, the action items, the project ideas, they're all in there but nobody's going back to read them. This skill turns a backlog of Granola transcripts into a connected Obsidian knowledge graph. Processes meetings in monthly batches, extracts the stuff that matters, and wires it all together with wiki-links. We used it to process ~600 meetings into something actually useful: [My Now Immaculate Knowledge Graph of Life](https://harper.blog/2026/03/11/2026-immaculate-knowledge-graph/)

![network-annotated_hu_dc5f11550bad1d72](https://github.com/user-attachments/assets/76a71ed6-018a-4dc6-8aff-79c2d76b86df)
*Obsidian graph view after processing ~600 meetings. Nodes are people and concepts, edges are co-occurrence in the same meeting.*

## Installation

```bash
/plugin marketplace add 2389-research/claude-plugins
/plugin install summarize-meetings@2389-research
```

## What it does

Point this at a month of meeting transcripts and it will:

- Read each transcript, skip the junk (empty stubs, scheduling fragments, recording setup chats)
- Dispatch 10-15 parallel agents to process the real meetings
- Write a summary file for each meeting with YAML frontmatter and a narrative recap
- Create or update People notes in `People/` (checks for existing entries and aliases first)
- Pull out concepts worth their own atomic note in `Concepts/`
- Stub out any projects mentioned that don't already exist
- Report back with a table of what got processed and what got skipped

## What gets extracted

From each meeting, agents pull out seven categories: people, action items, project ideas, blog ideas, knowledge graph connections, concepts, and general ideas. The summary file includes all of these plus a 2-4 paragraph narrative that reads like a real meeting recap — what happened, what mattered, what the vibe was.

## Triage

Not everything in `Meetings/transcripts/` is worth a summary. The skill skips files that are empty stubs, have fewer than 20 words of real content, are garbled recordings, or are medical appointments and kid-related content. Sparse but real notes (like bullet points from a fundraising call) still get processed.

## How it works

```
Scan month → Triage → Dispatch agents in waves → Compile report → Post update
```

Agents run in parallel (waves of 10-15) and each one handles a single transcript end-to-end. Large files (50K+) get chunked automatically.

## Documentation

The full workflow, templates, and extraction rules are in [skills/SKILL.md](skills/SKILL.md).

## Like this?

If Summarize Meetings saved you from transcript purgatory, a star ⭐️ helps us know it's landing.
````
