Agent skill

granola

This skill should be used when importing, listing, or exporting Granola meeting recordings and transcripts. Queries Granola's local cache and API to list meetings, extract transcripts, and export to Obsidian notes in Fathom-compatible format.

Stars 74
Forks 17

Install this agent skill to your Project

npx add-skill https://github.com/glebis/claude-skills/tree/main/granola

SKILL.md

Granola Meeting Importer

Query Granola's local data (cache + API) to list meetings, view transcripts, and export to Obsidian vault in the same format as the Fathom skill.

Prerequisites

  • Granola desktop app installed and authenticated (macOS)
  • No additional dependencies required (uses stdlib only)

Usage

bash
python3 ~/.claude/skills/granola/scripts/granola.py <command> [options]

Commands

Command Description
list List all meetings from local cache
show <id> Show meeting details (by ID prefix or title substring)
transcript <id> Get transcript (local cache, falls back to API)
export <id> Export meeting to Obsidian note (Fathom-compatible format)
api-list List meetings via Granola API (may show more than cache)

Options

Option Applies to Description
--format text|json list, transcript Output format (default: text)
--local-only transcript, export Skip API fallback, use only cached data
--vault <path> export Obsidian vault path (default: ~/Brains/brain)
--output <path> export Custom output file path
--limit <n> api-list Max results (default: 50)
--offset <n> api-list Pagination offset

Examples

List meetings

bash
python3 ~/.claude/skills/granola/scripts/granola.py list
python3 ~/.claude/skills/granola/scripts/granola.py list --format json

Export to Obsidian

bash
python3 ~/.claude/skills/granola/scripts/granola.py export bbeba240
python3 ~/.claude/skills/granola/scripts/granola.py export "Подкаст"

Get transcript

bash
python3 ~/.claude/skills/granola/scripts/granola.py transcript bbeba240
python3 ~/.claude/skills/granola/scripts/granola.py transcript "Подкаст" --format json

Output Format

Exported notes match Fathom skill format for consistency:

markdown
---
granola_id: <uuid>
title: "Meeting Title"
date: YYYY-MM-DD
participants: ['Name 1', 'Name 2']
duration: HH:MM
source: granola
---

# Meeting Title

## Summary
{AI-generated summary if available}

## Notes
{Markdown notes if available}

## Transcript
**Speaker Name**: What they said...

Files saved as: YYYYMMDD-meeting-title-slug.md

Data Sources

The script reads from two local sources:

  1. Cache file: ~/Library/Application Support/Granola/cache-v4.json -- metadata for all meetings, transcripts for active/recent meetings only
  2. API token: ~/Library/Application Support/Granola/supabase.json -- WorkOS bearer token for API calls (auto-refreshed when app is open, ~6h expiry)

See references/cache-structure.md for full schema documentation.

Auto-Sync (LaunchAgent)

The scripts/sync.sh script checks for new Granola meetings and exports any not yet in the vault. Designed to run on a schedule via macOS LaunchAgent.

How it works

  1. Scans existing vault .md files for granola_id: frontmatter to find already-exported meetings
  2. Calls granola.py api-list to get the 20 most recent meetings
  3. Exports any new ones via granola.py export
  4. Logs all activity to ~/Library/Logs/granola-sync.log

Setup

bash
# 1. Copy sync script (included in this skill)
chmod +x ~/.claude/skills/granola/scripts/sync.sh

# 2. Create LaunchAgent (runs every 15 minutes)
cat > ~/Library/LaunchAgents/com.user.granola-sync.plist << 'EOF'
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>Label</key>
    <string>com.user.granola-sync</string>
    <key>ProgramArguments</key>
    <array>
        <string>/bin/bash</string>
        <string>/Users/YOUR_USERNAME/.claude/skills/granola/scripts/sync.sh</string>
    </array>
    <key>StartInterval</key>
    <integer>900</integer>
    <key>StandardOutPath</key>
    <string>/Users/YOUR_USERNAME/Library/Logs/granola-sync.log</string>
    <key>StandardErrorPath</key>
    <string>/Users/YOUR_USERNAME/Library/Logs/granola-sync.log</string>
    <key>RunAtLoad</key>
    <false/>
</dict>
</plist>
EOF

# 3. Edit the plist: replace YOUR_USERNAME with your macOS username

# 4. Load the agent
launchctl load ~/Library/LaunchAgents/com.user.granola-sync.plist

# 5. Test manually
bash ~/.claude/skills/granola/scripts/sync.sh

Configuration

Set GRANOLA_VAULT environment variable to change the target vault (default: ~/Brains/brain).

bash
GRANOLA_VAULT=~/Obsidian/MyVault bash ~/.claude/skills/granola/scripts/sync.sh

Verify

bash
# Check agent is loaded
launchctl list | grep granola

# Check logs
tail -20 ~/Library/Logs/granola-sync.log

Known Limitations

  • Transcript content is sparse in cache -- only the most recent/active meeting typically has a full transcript locally. Older ones require API fetch.
  • No per-utterance speaker names -- Granola only provides source (microphone vs system audio). The export assigns the meeting creator to microphone utterances and "Other" to system audio.
  • Notes/summaries often empty -- Granola stores rich content server-side; the local cache has stubs. The API also returns empty for personal (non-workspace) docs.
  • Token expiry -- if the Granola app hasn't been open recently, the token may be expired. Open the app to refresh.

Integration

  • transcript-analyzer: After export, run transcript-analyzer on the output file for deeper analysis
  • Fathom skill: Granola exports use the same frontmatter and transcript format as Fathom exports, so downstream tools work with both

Didn't find tool you were looking for?

Be as detailed as possible for better results