From twinmind-pack
Implements TwinMind workflow for real-time/file-based meeting transcription with speaker diarization, AI summaries, action items, and follow-up emails using Ear-3 model.
npx claudepluginhub jeremylongshore/claude-code-plugins-plus-skills --plugin twinmind-packThis skill is limited to using the following tools:
- [Overview](#overview)
Guides testing TwinMind extension: transcribe sample meeting, generate AI summary/action items. For setup verification and basic patterns.
Provides reference architecture for Fireflies.ai meeting intelligence: GraphQL API client, webhook transcript processing, action items, analytics, CRM sync.
Transcribes audio/video to Markdown docs with LLM summaries, speaker diarization, timestamps, metadata using Faster-Whisper/Whisper. For meetings, interviews, lectures.
Share bugs, ideas, or general feedback.
Primary workflow for capturing meetings, generating transcripts with speaker diarization, and creating AI summaries with action items.
twinmind-install-auth setupBuild a MeetingCapture class with startLiveCapture() for real-time recording and transcribeRecording() for file-based transcription. Use Ear-3 model with auto language detection and speaker diarization.
Create a SummaryGenerator with generateSummary() (brief/detailed/bullet-points formats), generateFollowUpEmail(), and generateMeetingNotes() methods.
Build a SpeakerManager that extracts speakers from transcript segments, calculates speaking time per speaker, and optionally matches speakers to calendar attendees.
Wire everything together in processMeeting(): transcribe audio, then generate summary and identify speakers in parallel, optionally produce follow-up email and meeting notes.
See detailed implementation for complete MeetingCapture, SummaryGenerator, SpeakerManager, and orchestration code.
| Error | Cause | Solution |
|---|---|---|
| Transcription timeout | Large audio file | Increase maxWaitMs or use async callback |
| Speaker match failed | No calendar data | Provide attendees list manually |
| Summary generation failed | Transcript too short | Ensure minimum 30s of audio |
| Audio format unsupported | Wrong codec | Convert to MP3/WAV/M4A |
| Rate limit exceeded | Too many requests | Implement queue-based processing |
Basic usage: Apply twinmind core workflow a to a standard project setup with default configuration options.
Advanced scenario: Customize twinmind core workflow a for production environments with multiple constraints and team-specific requirements.
| Format | Supported | Notes |
|---|---|---|
| MP3 | Yes | Recommended |
| WAV | Yes | Best quality |
| M4A | Yes | iOS recordings |
| WebM | Yes | Browser recordings |
For action item extraction and follow-up automation, see twinmind-core-workflow-b.