MCP server

AI agents and coding assistants work with your videos: find one, import it, caption it, cut it, go live.

Call a tooljs
import { Client } from "@modelcontextprotocol/sdk/client/index.js";import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js"; // A project's API token, as for the APIconst transport = new StreamableHTTPClientTransport(  new URL("https://your-crumbcast/api/v1/mcp"),  { requestInit: { headers: { Authorization: `Bearer ${process.env.CRUMBCAST_TOKEN}` } } },);const client = new Client({ name: "my-agent", version: "1.0.0" });await client.connect(transport); const videos = await client.callTool({ name: "list_videos", arguments: { search: "launch" } });

What you get

  • One URLStreamable HTTP for Claude Code, Cursor or an agent of your own.
  • Your API token's rulesIts project only, at its role: a viewer token looks without changing anything.
  • Real work, in the backgroundImport, caption, clip and trim videos, read analytics, start live streams.
  • Reasons it can readWhat the API refuses, a tool refuses too, with the same reason.

More for developers