M↓ View as Markdown ↗

Actian MCP Server for Analytics Engine (Self-Hosted)¶

This page describes how to run the MCP Server against an Analytics Engine instance that you host yourself, on premises or in your own cloud account. You deploy the server in a container, supply the database connection details, and manage the server lifecycle.

If Analytics Engine powers a warehouse on the Actian Analytics AI Platform, the MCP Server is configured for you. See SaaS.

Prerequisites¶

Before starting the server, ensure the following requirements are met:

Container Engine: Docker installed and running on the host machine.
Database credentials: Valid access for the Analytics Engine database.
Security files (optional): TLS certificate and key files for secure deployments.
OIDC provider (optional): Required if you are using OAuth authentication.

Configuration¶

The server runs as a Docker container. To configure the server, mount the conf.json file to the container at /app/conf.json.

Create the Configuration File¶

Create a file named conf.json in your working directory using the following structure:

{
  "driver": "{Ingres}",
  "server": "@<db-host>,tcp_ip,<installation_id>",
  "database": "<database_name>",
  "max_connections": 10,
  "max_rows": 1000,
  "host": "<mcp_server_host>",
  "port": 8000,
  "query_mode": "read-only",
  "write_confirmation": true,
  "database_user": "<database_user>",
  "database_password": "<database_password>",
  "log_level": "INFO",
  "ssl_certfile": "/app/server.crt",
  "ssl_keyfile": "/app/server.key",
  "oauth": {
    "FASTMCP_SERVER_AUTH_CONFIG_URL": "<oidc_discovery_url>",
    "FASTMCP_SERVER_AUTH_CLIENT_ID": "<client_id>",
    "FASTMCP_SERVER_AUTH_CLIENT_SECRET": "<client_secret>",
    "FASTMCP_SERVER_AUTH_BASE_URL": "<server_base_url>",
    "FASTMCP_SERVER_AUTH_AUDIENCE": "<audience>",
    "user_impersonation": true
  }
}

Configuration Reference¶

Required Fields

Field	Type	Description
`driver`	`string`	ODBC driver name used to connect to Analytics Engine
`server`	`string`	Host or connection target for the Analytics Engine database
`database`	`string`	Name of the database.
`max_connections`	`integer`	Maximum concurrent database connections in the pool
`host`	`string`	Host address that the MCP Server listens to in the container
`port`	`integer`	Port that the MCP Server listens to in the container
`database_user`	`string`	Database username
`database_password`	`string`	Database password

Optional fields

Field	Type	Default	Description
`max_rows`	`integer`	`1000`	Maximum number of rows returned in a single query response. A statement that matches more rows is truncated to this limit, and the response includes the `truncated` and `warning` fields.
`log_level`	`string`	`INFO`	Server log verbosity. Valid values are `DEBUG`, `INFO`, `WARNING`, `ERROR`, `CRITICAL`.
`ssl_certfile`	`string`	None	Path to the TLS certificate file. Add `/app/server.crt` in the container.
`ssl_keyfile`	`string`	None	Path to the TLS private key file. Add `/app/server.key` in the container.
`oauth`	`object`	None	OAuth configuration block for protected deployments. For more information, see OAuth configuration.
`query_mode`	`string`	`read-only`	Controls whether data-modifying SQL is permitted. Valid values are `read-only` and `read-write`. See Write support.
`write_confirmation`	`boolean`	`true`	Whether a write requires human approval before it runs. Set to `false` only for clients that cannot display the approval prompt. See Write support. Applies only when `query_mode` is `read-write`.
`extensions`	`array`	None	Extension modules to load, each an object with a required `module` and an optional `config`. For more information, see Extensions.

Start the Server¶

Once you have created the conf.json file, start the container and mount the configuration file:

docker run -d \
    -v $(pwd)/conf.json:/app/conf.json:ro \
    -p 8000:8000 \
    --name=actian-mcp \
    actian/analytics-engine-mcp-server:1.1.0

Important

The container reads its configuration from /app/conf.json. Do not change the mount target path.

After the container starts, connect the MCP client to the server endpoint using the host and port specified in conf.json. For client configuration examples, see Connecting MCP Clients.

Usage¶

Once connected, the MCP client automatically discovers the server capabilities. You can perform the following tasks:

Inspect before querying: List tables and review structure before writing SQL.
Run a query: Execute a SQL statement and receive formatted results.
Explore functions: Look up available user-defined functions and stored procedures.

Next Steps¶

Write Support
Enable data-modifying SQL, and what gates each write.
Authentication
Secure the server with OAuth 2.0 and an external identity provider.
Extensions
Add custom tools to the server with a Python extension.