Skip to content

Latest commit

 

History

History
41 lines (30 loc) · 1.3 KB

File metadata and controls

41 lines (30 loc) · 1.3 KB

Transcript Library Agent Guide

This repository is a private internal tool for a small friend group. It is a proof of concept for sharing knowledge from a shared YouTube playlist. It is not a SaaS product.

What the app does

  • indexes videos and transcript parts from PLAYLIST_TRANSCRIPTS_REPO
  • renders a desktop-first app for browsing, watching, and reading
  • runs headless transcript analysis through local CLI providers
  • stores analysis artifacts under data/insights/<videoId>/

Non-goals

  • multi-tenant billing
  • public self-serve onboarding
  • end-user subscription management

Working rules

  • keep the app machine-keyed by videoId
  • also write human-readable slugged markdown files for manual review
  • keep provider switching behind the server runtime, not the UI
  • prefer additive migrations over destructive rewrites of stored artifacts
  • maintain observability through status.json, run.json, and worker logs

Runtime assumptions

  • this group already has access to Claude and ChatGPT/OpenAI tooling
  • the current deployment model is private and can rely on authenticated local or private worker CLIs
  • analysis execution should remain compatible with future worker separation

Primary commands

just start
just prod-start
just build
just lint
just typecheck
just backfill-insights