Ask an AI about this archive

A retrieval-augmented chat that searches the transcripts and answers with citations. It runs entirely in your browser and calls your own provider (Anthropic, OpenAI, or Google Gemini) with a key you supply — the key stays on your device and requests go straight to the provider.

This archive publishes a small, fixed set of discovery files. An agent (e.g. Claude Code via WebFetch) can read these and navigate every transcript without any per-video pages — the paginated shard scheme is documented inline.

• /llms.txt — an LLM-readable overview and link map.
• /corpus.json — the channel index and exactly how to fetch any transcript from the JSON shards.

The repo ships an MCP server that exposes this archive to Claude Code, Claude Desktop, Cursor, and other MCP clients as tools (search_transcripts, get_transcript, …). It reads the same static shards — over HTTP or from a local build. Add it to a client:

{
  "mcpServers": {
    "jeralyzer": {
      "command": "pnpm",
      "args": [
        "-C", "/path/to/yt-dlp-transcript-browser",
        "--filter", "yt-dlp-transcript-mcp",
        "exec", "tsx", "src/index.ts"
      ],
      "env": { "TRANSCRIPT_SITE_URL": "https://jeralyzer.pages.dev" }
    }
  }
}

Setup details and the claude mcp add command are in mcp/README.md.