TrackTag Launches a Music Tagging API, and an MCP Server for Claude & Cursor

Until today, TrackTag lived where you could see it: paste a track into the Lab, or run a batch through Studio. From today, it also lives everywhere your code, and your AI, lives. We're launching two things at once: the TrackTag API and the TrackTag MCP server.

The TrackTag API: auto-tag any catalog programmatically

The API gives developers, platforms and labels direct access to the same analysis engine behind Studio. Send a track, get back 35+ metadata fields, BPM, musical key, genres and subgenres, moods, emotions, instruments, vocal type, energy, danceability, a written description and a ready-to-paste keyword list.

It's built the way you'd hope a modern API is built:

• Async by default, POST a track, get a job id instantly, and we push the finished result to your webhook (signed, with automatic retries), or poll for it. Add ?wait=true for a one-round-trip synchronous call.
• No uploads to us, pass any URL (your CDN, an S3 signed link) or inline audio. We analyze and discard; nothing is stored except the results.
• Fair billing, the same credits you use in Studio, from $0.10 per track. Failed analyses are refunded automatically. Free test keys return sample responses so you can build without spending a cent.

A first integration is genuinely three lines of curl:

curl -X POST https://aaeabanvqnndwgrqsmhg.supabase.co/functions/v1/api-v1/v1/analyze \
  -H "Authorization: Bearer tt_live_YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"audio_url": "https://your-cdn.com/track.mp3", "model": "core"}'

Full reference, webhook signature verification snippets and error docs live at tracktag.me/developers. Keys are self-serve in Studio under the new API & MCP tab.

The MCP server: "analyze this track" inside Claude and Cursor

Here's the part we're most excited about. MCP (Model Context Protocol) is the standard that lets AI assistants use external tools, and TrackTag now speaks it natively. Add one block to your AI app's config, and your assistant learns four new skills: analyze a local audio file, analyze a URL, check your credit balance, and list recent analyses.

In practice it feels like magic: drag an MP3 into Claude Desktop, type "analyze this track with TrackTag", and watch the full metadata sheet come back in the conversation, BPM, key, genre, moods, instruments, the lot. Your assistant can even tell you how many credits you have left.

"tracktag": {
  "command": "npx",
  "args": ["-y", "tracktag-mcp"],
  "env": { "TRACKTAG_API_KEY": "tt_live_YOUR_KEY" }
}

The server is open on npm, listed in the official MCP Registry, and runs entirely on your machine, your audio and your key never pass through anyone else's servers on the way to TrackTag. As far as we know, TrackTag is the first music-tagging platform with a native MCP server.

Who is this for?

• Sample libraries & beat marketplaces, auto-tag every upload the moment it lands.
• Labels & distributors, enrich delivery spreadsheets for thousands of tracks in one scripted pass.
• Indie developers, add professional audio analysis to a DJ tool or plugin without training a model.
• Anyone with an AI assistant, tag your own music by dragging files into a chat.

Get started in five minutes

1. Open TrackTag Studio → API & MCP tab.
2. Create a key (every account works; credits are shared with Studio).
3. Run the quickstart curl, or paste the MCP config into Claude Desktop.

The developer docs cover everything else, webhooks, rate limits, test mode and errors. We'd love to see what you build: tell us at support and we'll happily bump your rate limits.