Documentation Index

Fetch the complete documentation index at: https://helpcenter.autify.com/llms.txt

Use this file to discover all available pages before exploring further.

Connecting to the Nexus MCP Server

Prev Next

This article walks you through connecting an MCP-compatible AI client such as Claude Desktop, Cursor, or VS Code to the Autify Nexus MCP Server. For an overview of the MCP Server and the tools it provides, see Nexus MCP Server.

Prerequisites

  • You are using a Nexus remote workspace.

  • You have an account with permission to issue API clients (typically a workspace administrator).

  • An MCP-compatible AI client (Claude Desktop, Cursor, VS Code, etc.) is installed.

Step 1: Obtain an API client secret

The MCP Server uses your Nexus API client secret as a bearer token. For instructions on creating an API client and retrieving its secret, see Obtain the client secret.

Step 2: Find your MCP endpoint URL

The MCP Server URL is your workspace's server URL with /nexus/mcp appended.

  1. Open Settings by clicking the gear icon at the bottom-left of the desktop app.

  2. Copy the Server URL shown in the Account section.

  3. Append /nexus/mcp to the URL you copied. The result is your MCP endpoint.
    Example: https://<organization-id>.cloud.autify.com/nexus/mcp

In the steps that follow, <your-org-id> stands for the subdomain part of your server URL, and <your-client-secret> stands for the API client secret you obtained in Step 1. Replace them with your own values when you set up your client.

Step 3: Configure your AI client

Register the MCP Server URL and credentials in your AI client's configuration file.

Claude Desktop

Configuration file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

From the Claude Desktop menu, choose SettingsDeveloperEdit Config to open (or create) the configuration file. Add the following entry:

{
  "mcpServers": {
    "nexus": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://<your-org-id>.cloud.autify.com/nexus/mcp",
        "--header",
        "Authorization: Bearer <your-client-secret>"
      ]
    }
  }
}

Note:

  • After saving the configuration, restart the application.

Cursor

Configuration file:

  • Per project: .cursor/mcp.json in the project root

  • Per user (shared across all projects): ~/.cursor/mcp.json

You can also add the server through the GUI by choosing Cursor SettingsMCPAdd new MCP server. Add the following entry:

{
  "mcpServers": {
    "nexus": {
      "url": "https://<your-org-id>.cloud.autify.com/nexus/mcp",
      "headers": {
        "Authorization": "Bearer <your-client-secret>"
      }
    }
  }
}

VS Code

Configuration file:

  • Per workspace: .vscode/mcp.json in the workspace root

  • Per user: the "mcp" section in your user settings.json

You can also add the server from the Command Palette (Cmd/Ctrl + Shift + P) by running MCP: Add Server. Add the following entry:

{
  "mcp": {
    "servers": {
      "nexus": {
        "type": "http",
        "url": "https://<your-org-id>.cloud.autify.com/nexus/mcp",
        "headers": {
          "Authorization": "Bearer <your-client-secret>"
        }
      }
    }
  }
}

Note:

  • To use MCP servers from VS Code, you need the GitHub Copilot Chat extension (or another extension with MCP support) installed.

Step 4: Verify the connection

Once the connection is established, Nexus's MCP tools become available in your AI client. The read-only list_workspaces tool is a convenient way to confirm that things are working. Try asking your AI client something like:

"Run the list_workspaces tool on Nexus and show me the workspaces I can access."

If you receive a list of workspaces in response, the connection is working correctly.

Limitations

  • Rate limit: Up to 10 requests per second per IP address. Exceeding this limit returns HTTP 429.

  • Mobile app testing tools are available only when your license includes them.

  • Connecting to a local workspace via MCP is not supported.

Troubleshooting

Symptom

Likely cause

Resolution

HTTP 403 "Invalid Request"

The host in the request URL does not match the subdomain of the organization that owns the API client.

Verify that the <your-org-id> in your URL matches the subdomain of the Server URL you confirmed in Step 2.

HTTP 429

You exceeded the rate limit (10 requests per second per IP).

Reduce how often your AI client sends requests.

Authentication error (401, etc.)

The secret is incorrect, or the API client has been deleted.

Open the "Manage API clients" screen, confirm the client still exists, and regenerate the secret if necessary.

The server fails to start in Claude Desktop

The npx command cannot be found (Node.js is not installed).

Install Node.js (the latest LTS version is recommended) and confirm that npx --version runs successfully in your terminal.