cato

The Cato MCP CMA server implements the Model Context Protocol to expose Cato's GraphQL API through a standardized MCP interface. It provides a suite of tools that let AI agents query entities, sites, users, and site metrics, as well as retrieve real-time WAN health, network statistics, and user/session details. The available tools are organized into categories such as Entity Lookup, Sites, Users and Clients, Site Metrics, and User Metrics. Clients like Claude Desktop or Cursor can connect to the server via an MCP client configuration (for example using a Docker-based deployment) and request data through these tools. When deployed with Docker, the server runs as ghcr.io/catonetworks/cato-mcp-server:latest and expects Cato authentication environment variables to be provided (CATO_API_HOST, CATO_ACCOUNT_ID, CATO_API_KEY).

To use the tools, configure an MCP client with the provided server name (e.g., cato) and point it at the running container. The client can then issue standardized MCP requests such as entity_lookup, site_details, user_details, site_metrics, and top_site_bandwidth_consumers to retrieve structured data from Cato's GraphQL API. The server handles authentication, data shaping, and pagination where applicable, returning results in a consistent MCP response format for downstream agents to parse and reason about.

Prerequisites:

• Docker installed on the host
• Access to ghcr.io to pull the image ghcr.io/catonetworks/cato-mcp-server:latest
• Cato API credentials (host, account ID, API key)

Step 1: Prepare environment variables

• Obtain your CATO_API_HOST, CATO_ACCOUNT_ID, and CATO_API_KEY from your Cato account or admin.

Step 2: Run the MCP server via Docker

• Use the following command to start the server (adjust environment values as needed):

docker run --rm --pull always -i \
  -e CATO_API_HOST=<your Cato API Host> \
  -e CATO_ACCOUNT_ID=<your Cato Account ID> \
  -e CATO_API_KEY=<your Cato API Key> \
  ghcr.io/catonetworks/cato-mcp-server:latest

Step 3: Verify the server is running

• Ensure your container is running and accessible from the MCP client. You should see the server listen on the default MCP interface used by your client (often via localhost:port or a bridged network).

Step 4: Configure your MCP client

• In Claude-Desktop or Cursor, add an MCP server entry named "cato" and point to the running container. Use the configuration example in the README to supply environment variables if needed.

Tips and considerations:

• The Docker image uses --pull always to fetch the latest version. If you want to freeze a version, remove --pull always and manually pull when you decide to upgrade.
• Ensure network access between the MCP client and the Docker host where the container runs.
• Provide CATO_API_HOST, CATO_ACCOUNT_ID, and CATO_API_KEY as environment variables; these are required for authenticating to Cato’s GraphQL API.
• The server exposes a rich set of tools; start with entity_lookup, site_details, user_details, and site_metrics to validate connectivity before exploring more advanced metrics.
• If you encounter permission prompts or access scopes in your MCP client, verify that the container has appropriate network access and that the client is allowed to request data from the server.