Cato MCP CMA
@catonetworks
This repository provides a local mcp-server for Cato Network's public GraphQL API
Overview
What is Cato MCP CMA?
Cato MCP CMA is a Model Context Protocol (MCP) server that integrates with the Cato CMA Public API. It allows AI assistants to query and retrieve information from Cato systems via GraphQL in a standardized way. Designed for use with MCP clients like Cursor and Claude Desktop (Claude Sonnet 4 recommended).
How to use Cato MCP CMA?
Run the server as a Docker container from ghcr.io/catonetworks/cato-mcp-server or build from source using yarn install && yarn build. Configure the required environment variables (CATO_API_HOST, CATO_ACCOUNT_ID, CATO_API_KEY) and add the server to your MCP client’s configuration (e.g., Claude Desktop’s claude_desktop_config.json).
Key features of Cato MCP CMA
- Lookup entities (users, sites, services) with filtering and pagination.
- Retrieve site details, location, connectivity, and socket versions.
- Access user profile, connection status, and software versions.
- Query site network health and bandwidth consumption rankings.
- Analyze user and site metrics with timeseries and aggregation.
Use cases of Cato MCP CMA
- AI assistants answering questions about site operational status and connectivity.
- Monitoring user connection details and session data.
- Identifying sites with poor network quality (high packet loss, latency, jitter).
- Capacity planning using top bandwidth consumers among sites or users.
- Analyzing infrastructure change events and annotation trends.
FAQ from Cato MCP CMA
What clients are compatible with Cato MCP CMA?
Tested with Cursor and Claude Desktop (non-free tier) using the Claude Sonnet 4 model.
What environment variables are required?
CATO_API_HOST (e.g., api.catonetworks.com), CATO_ACCOUNT_ID, and CATO_API_KEY for authentication.
How do I update to the latest version?
Use --pull always in the Docker run command, or manually run docker pull ghcr.io/catonetworks/cato-mcp-server:latest.
Can I build from source?
Yes, clone the repository, run yarn install and yarn build, then configure the MCP client to use the built index.js.
Does this server store any data locally?
No, it queries the Cato CMA Public API on each request; no local data persistence is mentioned.