MCP.so
Sign In
I

Iceberg MCP

@ryft-io

About Iceberg MCP

IcebergMCP is an MCP server that lets you interact with your Apache Iceberg Lakehouse using natural language in Claude, Cursor, or any other MCP client.

Basic information

Category

Other

Transports

stdio

Publisher

ryft-io

Submitted by

yguy-ryft

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "iceberg-mcp": {
      "command": "uv",
      "args": [
        "run",
        "--with",
        "iceberg-mcp",
        "iceberg-mcp"
      ],
      "env": {
        "ICEBERG_MCP_PROFILE": "<aws-profile-name>"
      }
    }
  }
}

Tools

5

Gets all namespaces in the Iceberg catalog

Gets all tables for a given namespace

Returns the schema for a given table

Returns table properties for a given table, like total size and record count

Gets all partitions for a given table

Overview

What is Iceberg MCP?

Iceberg MCP is a Model Context Protocol (MCP) server that lets you interact with an Apache Iceberg Lakehouse using natural language in Claude, Cursor, or any other MCP client. It connects to a catalog managed in AWS Glue and exposes table metadata as read-only tools.

How to use Iceberg MCP?

Install the uv package manager, then configure the server in your MCP client (Claude, Cursor) by adding a JSON entry with the command uv run --with iceberg-mcp iceberg-mcp and the environment variable ICEBERG_MCP_PROFILE set to your AWS profile name. Optionally set ICEBERG_MCP_REGION (default us-east-1).

Key features of Iceberg MCP

  • List all namespaces in the Iceberg catalog
  • List all tables for a given namespace
  • Return the schema for a specific table
  • Show table properties like total size and record count
  • Get all partitions for a given table

Use cases of Iceberg MCP

  • Discover namespaces and tables in your lakehouse by asking in natural language
  • Quickly inspect a table’s schema or partition layout
  • Retrieve table size and row counts without writing SQL
  • Generate SQL queries based on table metadata (e.g., aggregations on numeric columns)
  • Diagnose performance changes (e.g., “Why did queries on raw_events become slower?”)

FAQ from Iceberg MCP

What catalogs does Iceberg MCP support?

Currently only AWS Glue. Iceberg REST Catalog support is coming soon.

Can Iceberg MCP modify or delete data?

No. All tools are read-only and cannot modify or delete any data in your lakehouse.

What AWS configuration is required?

You need an AWS profile configured on the machine with access to the Glue catalog. Use the ICEBERG_MCP_PROFILE environment variable to specify the profile; ICEBERG_MCP_REGION defaults to us-east-1.

What dependencies does Iceberg MCP need?

The uv package manager must be installed. The server runs as a Python package via uv run.

Where does my data live?

Your data stays in your own AWS Glue catalog and object storage. The server only reads metadata; no data is sent to third parties.

Comments

More Other MCP servers