pocket-ai
par asabovetech
Search & ResearchRésumé
**Voice recording transcription, semantic search, and meeting intelligence across all conversations.**.
Description
Ce que fait ce skill
Pocket AI is a wearable device that records meetings, calls, and conversations, then transcribes and indexes them. The API exposes semantic search across all recordings, plus endpoints for listing recordings, filtering by tags, and downloading audio.
Semantic search finds relevant moments by meaning across every recording, so a query like 'manufacturing restructuring decisions' returns results even when those exact words were never spoken.
Quand l'utiliser
- Searching what was discussed with a specific person last week
- Pulling action items from yesterday's team calls
- Finding what decisions were made about a project topic
- Reviewing conversation context before a follow-up meeting
- Surfacing follow-ups that were mentioned but never written down
Exemple de workflow
User asks: What action items came out of my meetings today?
- 1
Send POST to /public/search with query 'action items tasks follow up today'
- 2
Parse relevantMemories array for entries with action_item in the sources field
- 3
Extract content, recordingDate, and speakers from each matching memory
- 4
Format results as a list grouped by recording title
A list of action items with timestamps, speaker labels, and source recording dates
Prérequis
- Pocket AI account and paired wearable device (heypocketai.com)
- API key from the Pocket AI dashboard, stored at ~/.config/pocket-ai/api_key
Installation
clawhub install asabovetech/pocket-aiExécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.