steampipe-mcp-server

# Steampipe MCP Server An MCP server interacting with PostgreSQL databases, primarily for use with [Steampipe](https://steampipe.io/). Steampipe has a schema per connection, and creates a search_path that includes all the schemas, but public schema is typically empty. In additiona to that, Steampipe plugins for AWS, GCP and other clouds have a lot of tables, so just listing all of them is not practical. So, the recommended way to prompt your Claude Desktop would be to say something like this: "In steampipe using aws_all schema, give me a list of all ec2 instances". This way Claude will be more likely to use list_tables_in_schema in schema command, to limit the number of tables retrieved. ## Prerequisites ### 1. Install Steampipe **macOS:** ```bash brew tap turbot/tap brew install steampipe ``` **Linux:** ```bash sudo /bin/sh -c "$(curl -fsSL https://steampipe.io/install/steampipe.sh)" ``` **Windows:** ```bash iwr -useb https://steampipe.io/install/steampipe.ps1 | iex ``` ### 2. Start Steampipe Service Start Steampipe as a background service: ```bash steampipe service start ``` You can verify the service is running with: ```bash steampipe service status ``` ### 3. Get Database URL The Steampipe PostgreSQL connection string can be found: ```bash steampipe service status ``` Look for the `Database URL` in the output, which typically looks like: ``` postgres://steampipe:password@localhost:9193/steampipe ``` You can provide this URL in the `--database-url` argument when running the server: ``` steampipe-mcp-server --database-url postgresql://steampipe:password@localhost:9193/steampipe ``` Note: Protocol must be `postgresql://` for the server to work correctly. ### 4. Configuring Environment Variables You can configure the database connection using an environment variable instead of passing it each time: 1. Create a `.env` file in the project directory with your database URL: ``` STEAMPIPE_MCP_DATABASE_URL=postgresql://steampipe:password@…