MCP for AI Assistants
Archivarix Tube Search provides a Model Context Protocol (MCP) server that lets AI assistants search and analyze archived YouTube videos directly. Use it from Claude Desktop, Claude Code, Cursor, VS Code, or any MCP-compatible client.
What is MCP?
Model Context Protocol is an open standard that lets AI assistants connect to external data sources and tools. Instead of copying and pasting information, your AI assistant can directly search our archive of 1.5 billion+ YouTube videos.
What Can You Do?
- Search channels — find all known videos from any YouTube channel, including deleted ones
- Full-text search — search across video titles and descriptions
- Read video metadata — get titles, descriptions, dates, view counts, and status
- Download subtitles — get full transcripts as readable text
- AI summaries — generate and read AI-powered video summaries
- Track status changes — see when videos were deleted, restored, or changed status
Requirements
- An MCP + API subscription (see pricing)
- An API key (generated in Profile → API)
- An MCP-compatible AI client
Supported Clients
| Client | Transport | Setup |
|---|---|---|
| Claude Desktop | stdio (npm) or Remote URL | Quickstart |
| Claude Code | stdio (npm) or Remote URL | Quickstart |
| Cursor | stdio (npm) | Quickstart |
| VS Code (Copilot) | stdio (npm) | Quickstart |
| Any MCP client | Streamable HTTP | Auth docs |
Open Source
The MCP client is open-source under the MIT license. Source code, issue tracker, and contributions:
- github.com/Archivarix-com/tube-search-mcp — repository
- npmjs.com/package/tube-search-mcp — published package
- Official MCP Registry — published as
io.github.Archivarix-com/tube-search-mcp
The npm package (npx tube-search-mcp) is a thin stdio-to-HTTP proxy that connects to the remote MCP endpoint at https://tube.archivarix.net/mcp. The backend service is proprietary, but the client wrapper that ships to your machine is fully auditable.