Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Capture your first meeting with Granola and review AI-enhanced notes. Use when testing Granola setup, learning the notepad + transcript flow, or understanding how Enhance Notes and Granola Chat work. Trigger: "granola hello world", "first granola meeting", "try granola", "granola test".
.claude/skills/jeremylongshore-granola-hello-world/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 34% | 0% |
Capture your first meeting with Granola and learn the three-phase workflow: live notepad, Enhance Notes, and Granola Chat. Granola captures audio from your device (no bot joins), transcribes it, and merges your typed notes with the full transcript to produce structured output.
granola-install-auth setupDuring the meeting, type rough notes directly in the Granola notepad:
markdown- discussed Q1 priorities - sarah owns onboarding revamp - need design review by thursday - budget question - follow up with finance
These notes give Granola context. The AI uses your notes plus the full transcript to produce better output — you do not need to capture everything.
Granola produces structured notes based on your template:
markdown# Team Standup — March 22, 2026 ## Summary Discussed Q1 priorities. Agreed to focus on customer onboarding improvements. Sarah will lead the initiative with a design review by Thursday. ## Key Decisions - Onboarding revamp is the top priority for Q1 - Budget approval needed from Finance before expanding scope ## Action Items - [ ] @sarah: Schedule design review meeting by Thursday - [ ] @mike: Create onboarding improvement tickets in Linear - [ ] @alex: Follow up with Finance on budget allocation ## Participants Sarah Chen, Mike Johnson, Alex Kim
After enhancement, use the chat panel to ask questions about the meeting:
Granola Chat has full context of the transcript and your notes.
Type / in Granola Chat to invoke Recipes — saved prompts for common tasks:
/standup — extract blockers and progress updates/follow-up — draft a follow-up email/action-items — list all action items with owners| Error | Cause | Fix | |-------|-------|-----| | No transcript generated | Wrong audio device selected | Check System Settings > Sound > Input matches your meeting audio | | Meeting not detected | Calendar event missing video link | Add Zoom/Meet/Teams link to event, or manually start recording | | Processing stuck > 5 min | Connectivity issue | Check internet, restart Granola (right-click icon > Restart) | | Empty enhanced notes | Meeting < 2 minutes | Granola needs sufficient audio for meaningful enhancement | | Poor transcription | Background noise or crosstalk | Use headset, mute when not speaking |
After your first meeting, check the People and Companies views in the sidebar. Granola automatically:
Proceed to granola-core-workflow-a for meeting template setup and preparation workflows.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,175 | 13,461 | -11% | 1 | 1 | 0% | 2,036 | 3,092 | +52% | 0 | 0 | — |
case-02 | fail→pass | 11,654 | 11,478 | -2% | 1 | 1 | 0% | 2,260 | 2,497 | +10% | 0 | 0 | — |
case-03 | fail→pass | 12,502 | 6,060 | -52% | 1 | 1 | 0% | 2,315 | 2,427 | +5% | 0 | 0 | — |
case-04 | pass→pass | 12,046 | 2,740 | -77% | 1 | 1 | 0% | 1,290 | 1,647 | +28% | 0 | 0 | — |
case-05 | pass→pass | 12,187 | 6,597 | -46% | 1 | 1 | 0% | 1,378 | 1,419 | +3% | 0 | 0 | — |
case-06 | pass→pass | 11,332 | 8,419 | -26% | 1 | 1 | 0% | 1,568 | 2,263 | +44% | 0 | 0 | — |
case-07 | fail→pass | 10,059 | 8,434 | -16% | 1 | 1 | 0% | 1,577 | 1,693 | +7% | 0 | 0 | — |
case-08 | fail→pass | 16,567 | 7,835 | -53% | 1 | 1 | 0% | 1,747 | 1,609 | -8% | 0 | 0 | — |
case-09 | pass→pass | 16,801 | 4,331 | -74% | 1 | 1 | 0% | 1,762 | 1,823 | +3% | 0 | 0 | — |
case-10 | fail→pass | 12,933 | 7,672 | -41% | 1 | 1 | 0% | 1,166 | 1,557 | +34% | 0 | 0 | — |
case-11 | fail→pass | 13,615 | 7,235 | -47% | 1 | 1 | 0% | 1,366 | 1,476 | +8% | 0 | 0 | — |
case-12 | fail→pass | 15,160 | 8,071 | -47% | 1 | 1 | 0% | 1,633 | 1,628 | -0% | 0 | 0 | — |
case-13 | pass→pass | 7,800 | 7,702 | -1% | 1 | 1 | 0% | 1,164 | 1,552 | +33% | 0 | 0 | — |
case-14 | fail→pass | 5,427 | 7,199 | +33% | 1 | 1 | 0% | 944 | 1,472 | +56% | 0 | 0 | — |
case-15 | pass→pass | 11,986 | 7,359 | -39% | 1 | 1 | 0% | 1,117 | 1,475 | +32% | 0 | 0 | — |
case-16 | pass→pass | 10,123 | 2,804 | -72% | 1 | 1 | 0% | 1,598 | 1,575 | -1% | 0 | 0 | — |
case-17 | fail→pass | 11,996 | 7,199 | -40% | 1 | 1 | 0% | 1,089 | 1,495 | +37% | 0 | 0 | — |
case-18 | fail→pass | 10,165 | 2,168 | -79% | 1 | 1 | 0% | 787 | 1,431 | +82% | 0 | 0 | — |
case-19 | pass→pass | 15,407 | 3,597 | -77% | 1 | 1 | 0% | 1,794 | 1,669 | -7% | 0 | 0 | — |
case-20 | fail→pass | 9,614 | 6,865 | -29% | 1 | 1 | 0% | 722 | 1,408 | +95% | 0 | 0 | — |
case-21 | fail→fail | 17,419 | 13,328 | -23% | 1 | 1 | 0% | 2,085 | 2,421 | +16% | 0 | 0 | — |
case-22 | fail→fail | 22,006 | 14,229 | -35% | 1 | 1 | 0% | 2,579 | 3,718 | +44% | 0 | 0 | — |
case-23 | fail→fail | 15,832 | 11,085 | -30% | 1 | 1 | 0% | 3,037 | 3,050 | +0% | 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. The headline lift of +48 percentage points is the difference between those two pass rates over the 23 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.