Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Zapier automation patterns and Enterprise API integration for Granola. Use when building automated workflows, connecting Granola to 8,000+ apps via Zapier, or querying the Enterprise API for notes and transcripts. Trigger: "granola zapier", "granola automation", "granola API", "granola SDK".
.claude/skills/jeremylongshore-granola-sdk-patterns/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 84% | 0% |
Granola does not have a traditional SDK. Integration is achieved through three channels: Zapier (8,000+ app connections), the Enterprise API (REST, workspace-level read access), and native integrations (Slack, Notion, HubSpot, Attio, Affinity). This skill covers automation patterns for all three.
Granola provides two Zapier triggers:
| Trigger | Fires When | Use Case | |---------|-----------|----------| | Note Added to Granola Folder | A note is placed in a specific folder | Auto-route by meeting type | | Note Shared to Zapier | You manually share a note to Zapier | Selective sharing for important meetings |
Webhook payload data available:
title — meeting title from calendarcreator_name / creator_email — note creatorattendees[] — array of {name, email} objectscalendar_event_title — original calendar event namecalendar_event_datetime — meeting date/timenote_content — the enhanced note content (Markdown)Pattern 1: Meeting Notes to Notion (auto-archive)
yamlTrigger: Note Added to Granola Folder ("All Meetings") Action: Notion — Create Database Item Database: Meeting Archive Title: "{{title}}" Date: "{{calendar_event_datetime}}" Content: "{{note_content}}" Attendees: "{{attendees}}"
Pattern 2: Action Items to Asana/Linear
yamlTrigger: Note Shared to Zapier Filter: note_content contains "Action Items" Code Step (JavaScript): const lines = inputData.note_content.split('\n'); const actions = lines .filter(l => l.match(/^- \[ \]/)) .map(l => l.replace('- [ ] ', '')); output = actions.map(a => ({task: a})); Action: Linear — Create Issue (for each action) Title: "{{task}}" Team: Engineering Label: "meeting-action"
Pattern 3: Sales Call Summary to Slack + HubSpot
yamlTrigger: Note Added to Granola Folder ("Sales Calls") Path A — Slack: Action: Post Message to #sales-updates Message: | *New Sales Call:* {{title}} *Attendees:* {{attendees}} {{note_content}} [View full notes in Granola] Path B — HubSpot (via Zapier if not using native): Action: Find Contact by Email ({{attendees[0].email}}) Action: Create Engagement Note Body: "{{note_content}}"
Pattern 4: Meeting Follow-Up Email
yamlTrigger: Note Shared to Zapier Action: ChatGPT — Generate Follow-Up Email Prompt: "Write a professional follow-up email based on: {{note_content}}" Action: Gmail — Create Draft To: "{{attendees}}" Subject: "Follow-up: {{title}}" Body: "{{chatgpt_response}}" Action: Slack — Notify Message: "Follow-up draft ready for: {{title}}"
Available on Enterprise plan. API keys generated at Settings > API Keys (up to 5 per workspace).
bash# List all accessible notes (paginated) curl -s "https://api.granola.ai/v0/notes" \ -H "Authorization: Bearer $GRANOLA_API_KEY" \ -H "Content-Type: application/json" | jq '.notes[:3]' # Get a specific note with transcript curl -s "https://api.granola.ai/v0/notes/{note_id}" \ -H "Authorization: Bearer $GRANOLA_API_KEY" | jq '{title, summary, action_items}'
API characteristics:
Reverse-engineered endpoints (unofficial, for reference):
POST https://api.granola.ai/v2/get-documents # List documents (paginated)
POST https://api.granola.ai/v1/get-document-transcript # Get transcript
POST https://api.granola.ai/v1/get-workspaces # List workspaces
POST https://api.granola.ai/v1/get-documents-batch # Bulk fetch by IDsAuthentication uses WorkOS with refresh token rotation via POST https://api.workos.com/user_management/authenticate.
yamlName: Complete Meeting Follow-Up Pipeline Step 1 — Trigger: Granola: Note Added to Folder ("Client Meetings") Step 2 — Filter: Only continue if attendees contain external email domains Step 3 — Action: ChatGPT: Generate structured summary and follow-up email Step 4 — Action: Gmail: Create draft follow-up email to external attendees Step 5 — Action: Notion: Create page in Client Meeting Log database Step 6 — Action: Linear: Create issues from action items with "client" label Step 7 — Action: Slack: Post summary to #client-updates channel Step 8 — Action: HubSpot: Log meeting note on matched Contact/Deal
Organize Granola folders to drive different Zap behaviors:
| Folder | Zapier Trigger | Actions | |--------|---------------|---------| | Sales Calls | Auto | Slack #sales + HubSpot + follow-up email | | Engineering | Auto | Linear tasks + Notion wiki | | All Hands | Auto | Slack #general + Google Drive archive | | Interviews | Manual share | Greenhouse scorecard + hiring panel Slack | | 1-on-1s | None | Private, no automation |
| Error | Cause | Fix | |-------|-------|-----| | Zapier trigger not firing | Folder trigger misconfigured | Verify the exact folder name in Zapier matches Granola | | Missing note content | Note still processing | Add a 2-minute delay step at the start of the Zap | | API 429 Too Many Requests | Rate limit exceeded | Add delays between requests, implement backoff | | API 401 Unauthorized | Invalid or expired API key | Regenerate key at Settings > API Keys | | Attendee data empty | Calendar event has no attendee list | Add attendees to the calendar event |
Proceed to granola-common-errors for troubleshooting.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 18,591 | 13,996 | -25% | 1 | 1 | 0% | 2,863 | 3,770 | +32% | 0 | 0 | — |
case-02 | fail→pass | 14,366 | 11,130 | -23% | 1 | 1 | 0% | 2,875 | 4,365 | +52% | 0 | 0 | — |
case-03 | fail→pass | 19,032 | 11,844 | -38% | 1 | 1 | 0% | 3,092 | 3,789 | +23% | 0 | 0 | — |
case-04 | fail→pass | 11,245 | 10,398 | -8% | 1 | 1 | 0% | 1,877 | 2,645 | +41% | 0 | 0 | — |
case-05 | fail→pass | 12,541 | 8,772 | -30% | 1 | 1 | 0% | 1,324 | 2,431 | +84% | 0 | 0 | — |
case-06 | fail→pass | 25,541 | 9,068 | -64% | 1 | 1 | 0% | 1,437 | 2,515 | +75% | 0 | 0 | — |
case-07 | fail→pass | 14,598 | 8,768 | -40% | 1 | 1 | 0% | 1,664 | 2,426 | +46% | 0 | 0 | — |
case-08 | fail→fail | 20,139 | 17,963 | -11% | 1 | 1 | 0% | 2,527 | 4,023 | +59% | 0 | 0 | — |
case-09 | pass→pass | 19,580 | 7,394 | -62% | 1 | 1 | 0% | 2,518 | 3,058 | +21% | 0 | 0 | — |
case-10 | fail→pass | 21,614 | 5,907 | -73% | 1 | 1 | 0% | 1,364 | 2,754 | +102% | 0 | 0 | — |
case-11 | fail→pass | 18,075 | 7,003 | -61% | 1 | 1 | 0% | 2,174 | 2,060 | -5% | 0 | 0 | — |
case-12 | pass→pass | 13,349 | 18,439 | +38% | 1 | 1 | 0% | 2,458 | 4,142 | +69% | 0 | 0 | — |
case-13 | fail→fail | 19,886 | 14,611 | -27% | 1 | 1 | 0% | 2,334 | 3,936 | +69% | 0 | 0 | — |
case-14 | fail→pass | 10,521 | 8,293 | -21% | 1 | 1 | 0% | 1,595 | 2,362 | +48% | 0 | 0 | — |
case-15 | pass→pass | 20,852 | 17,111 | -18% | 1 | 1 | 0% | 2,461 | 3,629 | +47% | 0 | 0 | — |
case-16 | pass→pass | 14,858 | 3,941 | -73% | 1 | 1 | 0% | 1,658 | 2,384 | +44% | 0 | 0 | — |
case-17 | pass→pass | 11,707 | 11,621 | -1% | 1 | 1 | 0% | 1,863 | 2,880 | +55% | 0 | 0 | — |
case-18 | pass→pass | 12,135 | 10,110 | -17% | 1 | 1 | 0% | 1,879 | 2,649 | +41% | 0 | 0 | — |
case-19 | fail→pass | 12,304 | 2,238 | -82% | 1 | 1 | 0% | 1,910 | 2,031 | +6% | 0 | 0 | — |
case-20 | fail→pass | 9,073 | 4,355 | -52% | 1 | 1 | 0% | 1,595 | 2,526 | +58% | 0 | 0 | — |
case-21 | fail→fail | 9,112 | 15,672 | +72% | 1 | 1 | 0% | 1,638 | 3,215 | +96% | 0 | 0 | — |
case-22 | pass→pass | 18,358 | 17,089 | -7% | 1 | 1 | 0% | 2,606 | 3,962 | +52% | 0 | 0 | — |
case-23 | pass→pass | 14,366 | 8,550 | -40% | 1 | 1 | 0% | 1,849 | 3,095 | +67% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +52 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.