home / mcp / iceberg mcp server
MCP server for interacting with Apache Iceberg catalog from Claude, enabling data lake discovery and metadata search through a LLM prompt.
Configuration
View docs{
"mcpServers": {
"ahodroj-mcp-iceberg-service": {
"command": "uv",
"args": [
"--directory",
"PATH_TO_/mcp-iceberg-service",
"run",
"mcp-server-iceberg"
],
"env": {
"S3_ENDPOINT": "OPTIONAL IF USING S3",
"AWS_ACCESS_KEY_ID": "YOUR S3 ACCESS KEY",
"ICEBERG_WAREHOUSE": "YOUR ICEBERG WAREHOUSE NAME",
"ICEBERG_CATALOG_URI": "http://localhost:8181",
"AWS_SECRET_ACCESS_KEY": "YOUR S3 SECRET KEY"
}
}
}
}You can run an MCP Iceberg Catalog server to expose Apache Iceberg tables through a SQL interface inside Claude Desktop. This lets you list, describe, query, and insert data into Iceberg tables in a familiar SQL workflow while leveraging PyIceberg for table operations and PyArrow for data handling.
Connect Claude Desktop to the Iceberg MCP server you configured. Once connected, you can run standard SQL statements against Iceberg catalogs and tables. Supported operations include: LIST TABLES to enumerate tables, DESCRIBE TABLE to inspect table schemas, SELECT to read data, and INSERT to load data into Iceberg tables. The server translates SQL queries into Iceberg operations and returns results for you to view in Claude Desktop.
Prerequisites are needed before you install and run the MCP Iceberg Catalog server.
Install via Smithery to automatically wire the MCP into Claude Desktop.
Configuration uses environment variables to connect to your Iceberg REST catalog and storage. The server runs via a local command using the UV runtime and expects a specific directory path where the MCP service lives. You will set ICEBERG_CATALOG_URI, ICEBERG_WAREHOUSE, and optional S3 settings in your environment. The default Iceberg REST catalog URL is http://localhost:8181, and you provide your Iceberg warehouse name and optional S3 credentials.