Easily connecting AI services via MCP

In partnership with

Fellow AI enthusiasts,

Decades ago, programmers coded in assembly, close to the machine.

Then came higher-level languages like Python—more abstract, far more understandable.

Today, we're evolving again, starting to code with natural language itself. Think of it as continuous evolution, not distinct steps.

Guiding AI with language to generate code? That's "vibe coding." As vibe coders, we know the Model Context Protocol (MCP) is essential. This newsletter is about connecting to Elevenlabs via MCP, and how to bring AI agents into existence.

Further below, to our fellow Premium subscribers, there is also a video to follow on the implementation.

I’ve secured additional Manus access codes exclusively for my premium subscribers.

Build AI Agents (that do outbound calls) via the ElevenLabs MCP

Preparation - purchase a Twilio phone number:

1. Log in to your Twilio Console.

2. Navigate to Phone Numbers > Manage > Buy a number.

3. Search for available numbers using criteria like country, capabilities (Voice, SMS, etc.), and type.

4. Choose the number you want from the search results.

5. Click "Buy" and confirm the purchase (ensure you have sufficient funds or a payment method).

Preparation - important phone number to Elevenlabs:

Get Account SID and Auth Token from Twilio main page in dashboard, and import the phone number to Elevenlabs

Configure MCP in Cursor:

• Open Cursor's Settings.

• Navigate to the "MCP" section.

• Click to add a "New Global MCP Server".

• Paste the following JSON configuration into the settings:

  JSON

  {
    "mcpServers": {
      "ElevenLabs": {
        "command": "python",
        "args": ["-m", "elevenlabs_mcp.server"],
        "env": {
          "ELEVENLABS_API_KEY": "<YOUR_ELEVENLABS_API_KEY>" 
        }
      }
    }
  }

• Important: Replace <YOUR_ELEVENLABS_API_KEY> with your actual API key from ElevenLabs.

• (Alternatively, you can find this configuration structure in the ElevenLabs MCP GitHub repository).

2. Install uv and the ElevenLabs MCP Package:

• Open a new terminal within Cursor.

• Install the uv package manager by running:

  curl -LsSf https://astral.sh/uv/install.sh | sh 

3. Verify MCP Server Status:

• Go back to Cursor's Settings -> MCP section.

• You should now see the "ElevenLabs" server listed.

• Look for an indicator (like a green light or "running" status) confirming it's active.

• This running server provides functions like tools_to_speech, list_agents, create_agent, etc., for Cursor to use.

4. Use MCP via Prompt:

• With the server running, you can now instruct Cursor to use its capabilities.

• For example, open a chat or prompt window in Cursor and use a prompt like the one you provided:

• Note: Replace <RESTAURANT_PHONE_NUMBER> with the actual phone number.

• Cursor can now potentially use the ElevenLabs MCP functions (like create_agent) to process this request and generate the necessary code or agent configuration.

It works like a charm

See the video on how to implement it

Subscribe to Premium to See the Rest