Skip to main content

Connect MCP Clients

Once you have a deployed MCP Server, you can connect AI agents and assistants to it so they can discover and use your tools. This guide covers configuration for popular MCP clients and general setup for any MCP-compatible client.

Prerequisites​

Before connecting a client, ensure you have:

  • A deployed and active MCP Server. See Deploy an MCP Server for setup instructions.
  • Your MCP Server URL in the format https://api-genai.nexla.io/mcp/service_key/{your_server_key}.
  • Your Nexla service key for authentication (unless using OAuth).
tip

You can get a ready-to-paste configuration JSON from the export detail page or via the GET /v1/toolsets/{id}/mcp_config API endpoint. See Deploy an MCP Server for details.

Claude Desktop​

To connect Claude Desktop to your Nexla MCP Server:

  1. Open Claude Desktop and go to Settings.
  2. Navigate to the MCP Servers section (under the Developer tab if applicable).
  3. Add a new server with the following configuration:
{
  "mcpServers": {
    "nexla": {
      "url": "https://api-genai.nexla.io/mcp/service_key/{your_server_key}",
      "headers": {
        "Authorization": "Bearer YOUR_NEXLA_SERVICE_KEY"
      }
    }
  }
}
  1. Save the configuration and restart Claude Desktop.
  2. Your tools should now appear in the tool picker when starting a new conversation.
tip

Replace nexla with a descriptive name if you have multiple MCP servers connected, for example nexla-sales-tools or nexla-support-agent. The key name is a local label and does not affect server behavior.

Cursor​

To connect Cursor to your Nexla MCP Server:

  1. Open Cursor and go to Settings.
  2. Navigate to the MCP tab.
  3. Click Add MCP Server.
  4. Enter your MCP Server URL: https://api-genai.nexla.io/mcp/service_key/{your_server_key}.
  5. Configure the authentication header:
    • Header name: Authorization
    • Header value: Bearer YOUR_NEXLA_SERVICE_KEY
  6. Save and restart Cursor.

After restarting, Cursor will connect to the MCP server and make your tools available in AI-assisted coding sessions.

VS Code with Copilot​

To connect GitHub Copilot in VS Code to your Nexla MCP Server:

  1. Open VS Code and go to Settings (Ctrl/Cmd + ,).
  2. Search for "MCP" in the settings search bar.
  3. Add a new MCP server entry with your server URL and authentication headers:
    • URL: https://api-genai.nexla.io/mcp/service_key/{your_server_key}
    • Headers: Authorization: Bearer YOUR_NEXLA_SERVICE_KEY
  4. Save the settings.

Copilot will use the MCP server to access your tools when relevant to the coding context.

note

MCP support in VS Code Copilot may require specific VS Code and Copilot extension versions. Check the GitHub Copilot documentation for the latest requirements.

ChatGPT​

Nexla provides a dedicated ChatGPT integration that uses OAuth for authentication, allowing individual users to authorize access with their own credentials.

See Integrate Nexla with ChatGPT for complete setup instructions.

info

The ChatGPT integration uses OAuth rather than service key authentication. This means each user authorizes access individually, providing per-user audit trails and access control.

Other MCP Clients​

Any client that supports the MCP Streamable HTTP transport can connect to your Nexla MCP Server. The requirements are:

  • Support for the Streamable HTTP transport type
  • Ability to set custom HTTP headers (for service key authentication), or OAuth support (for OAuth authentication)

Use the following general configuration pattern:

{
  "url": "https://api-genai.nexla.io/mcp/service_key/{server_key}",
  "transport": "streamable-http",
  "headers": {
    "Authorization": "Bearer YOUR_NEXLA_SERVICE_KEY"
  }
}

Refer to your MCP client's documentation for the exact configuration format, as different clients may expect slightly different JSON structures.

Verifying the Connection​

After configuring your client, verify that the connection is working:

  1. Check tool discovery -- Your client should list the available tools from your ToolSet. Look for the tool names in the client's tool picker or tool list.
  2. Execute a test tool -- Ask the AI agent a question that requires data from one of your Nexsets. The agent should invoke the appropriate tool and return results.
  3. Review audit trail -- Check the Audit & Receipts page in the Nexla Platform to verify that the tool execution was recorded with the correct client and user information.

Troubleshooting​

Tools Not Appearing​

If your MCP client connects but no tools are listed:

  • Verify the MCP Server status is active (not draft, paused, or retired).
  • Confirm the service key is valid and has not been revoked.
  • Double-check the server URL -- ensure the server_key portion is correct.
  • Confirm your client supports the Streamable HTTP transport. Some older MCP clients only support SSE or stdio transports.
  • Try refreshing the tools cache from the export detail page. See Deploy an MCP Server for instructions.

Authentication Errors​

If you receive 401 Unauthorized or 403 Forbidden errors:

  • Verify your service key is correct and has not expired.
  • Ensure the Authorization header includes the Bearer prefix: Bearer YOUR_KEY, not just YOUR_KEY.
  • Check that the service key has permission to access the ToolSet associated with this MCP Server.
  • If using OAuth, verify the OAuth provider configuration in the Nexla Platform.

Connection Timeout​

If the client cannot reach the MCP Server:

  • Verify network access to api-genai.nexla.io from the machine running the MCP client.
  • Check whether a firewall or proxy is blocking outbound HTTPS connections.
  • Confirm the MCP Server is in active status.
  • Try accessing the server URL directly in a browser or with curl to rule out network issues:
curl -I https://api-genai.nexla.io/mcp/service_key/{your_server_key}

If issues persist, check the Nexla Platform for any service notifications, or contact Nexla support.