Model Context Protocol

Bring Pulse into your AI assistant.

Pulse runs as an MCP server, so any compatible assistant — Claude Desktop, Claude Code, Cursor, and others — can call it directly. Your assistant pays with your banked tokens; you keep one credential.

We'll stamp this into the snippets below as you type. Nothing is sent to a server — it lives only in this page. Don't have a key? Get one — it takes about ten seconds.

Heads up: the Pulse MCP currently accepts the API key only as a URL query parameter. Treat the URL as a secret — store it in a password manager, not in version control. Header-based bearer auth is on the v1.1 roadmap.

Claude Desktop

Open Settings → Developer → Edit Config, then merge this entry into the mcpServers object:

{
  "mcpServers": {
    "studiosphere-pulse": {
      "type": "http",
      "url": "https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY"
    }
  }
}

Restart Claude Desktop. The Pulse tools (estimate_cost, analyze_track, get_job_status, etc.) will appear in any conversation.

Claude Code (CLI)

One command, no manual JSON edits. The --transport http flag tells Claude Code this is a remote server, not a stdio binary:

claude mcp add --transport http studiosphere-pulse https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY

Run claude mcp list to verify. Pulse tools are available in your next session.

Cursor

Add this to ~/.cursor/mcp.json (create the file if it doesn't exist):

{
  "mcpServers": {
    "studiosphere-pulse": {
      "type": "http",
      "url": "https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY"
    }
  }
}

Cursor picks up the change on next launch.

VS Code / GitHub Copilot

Add Pulse as a user or workspace MCP server. For a workspace, create .vscode/mcp.json:

{
  "servers": {
    "studiosphere-pulse": {
      "type": "http",
      "url": "https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY"
    }
  }
}

Open Copilot Chat, choose tools, and enable StudioSphere Pulse for prompts that need BPM, key, or waveform analysis.

ChatGPT

OpenAI's MCP support is rolling out. The Pulse server speaks Streamable HTTP transport at the URL below — paste it into whichever connector / GPT-action surface your account has access to, and supply the API key as the api_key query parameter.

https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY

If your ChatGPT plan doesn't yet expose MCP, the same tools are available via the Pulse REST API directly — see /llms.txt for the agent-readable contract.

Raw transport details

For any MCP-aware client, the server uses Streamable HTTP transport:

The configured URL with your key embedded:

https://mcp.studiosphere.space/mcp?api_key=YOUR_API_KEY

Verify it's working

After restarting your client, ask the assistant: "What's the BPM of https://example.com/track.mp3?" The assistant should call the estimate_cost tool first, confirm the price, then call analyze_track and read the result. If the assistant doesn't see Pulse tools, double-check the URL has your key on it and that you restarted after editing the config.

Need the short version?

The public quickstart keeps the agent workflow, rights confirmation, and canonical marketplace links in one place: Pulse MCP quickstart.

What your assistant gets

estimate_costQuote a price for a public audio URL.
analyze_trackRun analysis (BPM, key, waveform). Auth required.
request_payment_linkStripe Checkout link for one-off use after rights confirmation.
get_job_statusPoll a running job, fetch the result.
get_token_balanceCheck banked-token balance.