zabbix-graphql-api/docs/howtos/mcp.md at 4ec61ffba171c75051647cad423fe9780b4ecbe6

Andreas Hilbig 4ec61ffba1 chore: add MCP integration and refactor documentation into modular how-to guides

- Moved GraphQL query samples into a new `docs/queries` directory for better organization.
- Added new queries and mutations, including `createHost.graphql` and `GetApiVersion.graphql`.
- Introduced `mcp-config.yaml` and updated `docker-compose.yml` for MCP integration.
- Updated IntelliJ `.idea/workspace.xml` settings to reflect project changes.
- Added new how-to guides (`docs/howtos`) for permissions, tags, MCP integration, and schema usage.
- Enhanced tests by updating file paths and improving sample data locations.
- Refined permissions and host group structures in `zabbix-hostgroups.ts` and `resolvers.ts`.

2026-01-30 00:47:02 +01:00

2.8 KiB

Raw Blame History

🤖 Model Context Protocol (MCP) Integration

The Zabbix GraphQL API supports the Model Context Protocol (MCP), enabling Large Language Models (LLMs) to interact directly with your Zabbix data through a standardized interface.

Overview

By leveraging GraphQL, the API provides a strongly-typed and introspectable interface that is ideal for MCP. This allows LLMs to:

Discover available queries and mutations.
Understand the data structures (hosts, items, templates, etc.).
Execute operations to retrieve or modify Zabbix data based on natural language prompts.

Running Apollo MCP Server with Docker Compose

You can start both the Zabbix GraphQL API and the Apollo MCP Server using Docker Compose. This setup uses a local mcp-config.yaml and a generated schema.graphql.

Prerequisites: Ensure you have a .env file with the required Zabbix connection details.
Generate Schema: Generate the combined schema file required by the MCP server:
```
cat schema/*.graphql > schema.graphql
```
Prepare Operations: Create the operations directory if it doesn't exist:
```
mkdir -p mcp/operations
```
Start Services:
```
docker-compose up -d
```

This will:

Start the zabbix-graphql-api on http://localhost:4001/graphql (internal port 4000).
Start the apollo-mcp-server on http://localhost:3000/mcp (mapped from internal port 8000), configured to connect to the local API via mcp-config.yaml.

Using with Claude Desktop

To use this integration with Claude Desktop, add the following configuration to your Claude Desktop config file (typically appflowy.json or similar depending on OS, but usually claude_desktop_config.json):

{
  "mcpServers": {
    "zabbix-graphql": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-v", "/path/to/your/project/mcp-config.yaml:/mcp-config.yaml",
        "-v", "/path/to/your/project/schema.graphql:/schema.graphql",
        "-v", "/path/to/your/project/mcp/operations:/mcp/operations",
        "-e", "APOLLO_GRAPH_REF=local@main",
        "ghcr.io/apollographql/apollo-mcp-server:latest",
        "/mcp-config.yaml"
      ]
    }
  }
}

Note: Ensure the zabbix-graphql-api is running and accessible. If running locally, you might need to use host.docker.internal:4001/graphql in your mcp-config.yaml to allow the containerized MCP server to reach your host.

Benefits of GraphQL-enabled MCP

Self-Documenting: The GraphQL schema provides all necessary metadata for the LLM to understand how to use the tools.
Efficient: LLMs can request exactly the data they need, reducing token usage and improving response speed.
Secure: Uses the same authentication and permission model as the rest of the API.

2.8 KiB Raw Blame History