Connect Claude to Transcodely

Transcodely ships a hosted MCP (Model Context Protocol) server that exposes a curated set of Transcodely tools to Claude and any other MCP-compatible client. It runs at:

https://mcp.transcodely.com/mcp

There is nothing to install and no gateway to run — the server speaks MCP over streamable HTTP and authenticates each request with a Transcodely API key. Point a client at the URL, attach your key, and Claude can submit jobs, ingest videos, and read back status in plain language.

MCP is an open standard, so the same URL and header work in Claude Code, claude.ai, and any third-party client that speaks streamable-HTTP MCP.

Before you start

You need an app-scoped API key (ak_…). Create or copy one from the dashboard — see API Keys for how keys map to apps, and Authentication for the key format. The MCP server accepts the same keys as the REST API; every tool call is scoped to the key’s app.

Add it to Claude Code

claude mcp add --transport http transcodely https://mcp.transcodely.com/mcp --header "Authorization: Bearer ak_..."

To share the connection with a repository (so teammates and CI pick it up), commit the equivalent entry in a .mcp.json at the project root:

{
  "mcpServers": {
    "transcodely": {
      "type": "http",
      "url": "https://mcp.transcodely.com/mcp",
      "headers": {
        "Authorization": "Bearer ak_..."
      }
    }
  }
}

Note: Treat .mcp.json like any other file that carries a credential. Prefer an environment reference or a per-developer key over committing a live ak_… value.

Add it to claude.ai

In claude.ai, open Settings → Connectors → Add custom connector and provide:

URL — https://mcp.transcodely.com/mcp
Authorization header — add a request header named Authorization with the full value Bearer ak_... (type the whole Bearer prefix followed by your key).

Custom connectors that authenticate with a request header are rolling out across claude.ai plans, so the header field may not be available on every workspace yet. If your account doesn’t expose it, connect through Claude Code (above) in the meantime — it talks to the same server.

Tools

The server exposes a focused set of tools that wrap the Transcodely API. Claude picks the right one from your request; you don’t call them directly.

Tool	What it does
`create_job`	Submit a transcoding job for an input on one of your origins.
`create_video_from_url`	Ingest a remote file into hosted storage and start processing it.
`get_video`	Look up a hosted video — details, renditions, and playback URLs.
`get_usage`	Return a summary of your monthly usage and cost.
`list_jobs`	List recent jobs, optionally filtered by status.
`generate_captions`	Queue caption generation for a hosted video (requires the captions feature).
`get_job_status`	Return compact progress and status for a single job.

Authentication and scoping

Every request to the MCP server is authenticated with the API key you attached, exactly like a REST call. The tools act as that key: they can see and change only the resources that belong to the key’s app — there is no cross-app or cross-organization access, and no tool can escalate beyond what the key already permits.

Your API key is a secret. Anyone holding it can submit billable jobs and read your videos, so store it like a password and never paste it into a shared prompt. Rotate or revoke keys any time from the dashboard; a revoked key stops working on the MCP server immediately. See API Keys for the full lifecycle.

Try it

Once the connector is attached, ask Claude in plain language:

Transcode https://example.com/input.mp4 to HLS 1080p.

Claude picks the matching tool — create_video_from_url or create_job — runs it against your app, and reports back the resulting ID and status. Follow up with “what’s the status of that job?” and it calls get_job_status, or “how much have I used this month?” to call get_usage.

#Connect Claude to Transcodely

#Before you start

#Add it to Claude Code

#Add it to claude.ai

#Tools

#Authentication and scoping

#Try it