Cortex

Cortex MCP is a server that provides access to the Cortex API by using context from your workspace. It helps you ask natural-language questions and get informed answers that reflect your system’s structure, making incident handling and planning faster and more accurate.

How to use

You connect your MCP client to the Cortex MCP server to start querying context-aware information. Use natural language prompts to ask questions like who should handle an incident, what services belong to a team, or what information is needed to handle an ongoing issue. The server processes your workspace context and returns relevant results to guide your decisions.

How to install

Prerequisites: ensure Docker is installed and running on your system. You will also need an MCP-compatible client (such as Claude Desktop, Cursor, or similar) and a Cortex Personal Access Token.

1. Prepare your Cortex API token in your Cortex workspace settings and note the token value. Replace YOUR_API_TOKEN_HERE with your actual token in the commands below.

2. Add Cortex MCP to your MCP client configuration. Use the following example configurations for the supported clients. Each snippet shows how to run the Cortex MCP server in Docker with the required token passed as an environment variable.

{
  "mcpServers": {
    "cortex": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "--env",
        "CORTEX_API_TOKEN=YOUR_API_TOKEN_HERE",
        "ghcr.io/cortexapps/cortex-mcp:latest"
      ]
    }
  }
}

claude mcp add-json "cortex" '{
  "command": "docker",
  "args": [
    "run",
    "--rm",
    "-i",
    "--env",
    "CORTEX_API_TOKEN=YOUR_API_TOKEN_HERE",
    "ghcr.io/cortexapps/cortex-mcp:latest"
  ]
}'

{
  "mcpServers": {
    "cortex": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "--env",
        "CORTEX_API_TOKEN=YOUR_API_TOKEN_HERE",
        "ghcr.io/cortexapps/cortex-mcp:latest"
      ]
    }
  }
}

{
  "inputs": [
    {
      "type": "promptString",
      "id": "cortex-key",
      "description": "Cortex API Key",
      "password": true
    }
  ],
  "servers": {
    "Cortex": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "ghcr.io/cortexapps/cortex-mcp:latest"
      ],
      "env": {
        "CORTEX_API_TOKEN": "${input:cortex-key}"
      }
    }
  }
}

{
  "cortex": {
    "command": "docker",
    "args": [
      "run",
      "--rm",
      "-i",
      "--env",
      "CORTEX_API_TOKEN=YOUR_API_TOKEN_HERE",
      "ghcr.io/cortexapps/cortex-mcp:latest"
    ],
    "env": {}
  }
}

Additional installation notes

If you are self-managed, set CORTEX_API_BASE_URL=https:// alongside the CORTEX_API_TOKEN variable in your configurations. This ensures the MCP client talks to the correct Cortex API endpoint.

Configuration, security, and notes

Keep your Cortex API token secure. Do not share it in public configurations. Use your MCP client’s secure storage for credentials whenever possible. If you encounter issues, verify Docker is running, the token is valid, and the environment variable is correctly passed to the container.

Troubleshooting

Common issues include Docker not running, invalid tokens, or misconfigured environment variables. Check that the container starts with the token in the environment and that your MCP client can reach the Cortex MCP server endpoint.