Marmot Data Catalog — Claude Code MCP server: data metadata
Gives your AI agent real metadata from the Marmot catalog: asset search, lineage, ownership. Useful when the agent writes code against tables and topics instead of guessing schemas.
Open on GitHubWhat Marmot Data Catalog does
Marmot is an open-source data catalog that ships as a single binary. It holds tables, topics, queues, APIs and dashboards with descriptions, owners and the links between them. This MCP server exposes the catalog so the agent reads certified context rather than guessing at structure.
The connection runs over HTTP at https://{marmot_host}/api/v1/mcp, where {marmot_host} is your deployed Marmot host. In Claude one command is enough: claude mcp add --transport http marmot-data-catalog https://{marmot_host}/api/v1/mcp. The server does nothing on its own — you need a running Marmot instance first, or there is nothing to connect to.
Through the agent you can search assets by name and metadata, trace where data comes from, and find who owns a table. MIT licensed, self-hosting is free.
What it helps with
- find a table by name and metadata without scrolling the catalog
- trace data lineage from source to destination
- check who owns an asset before changing its schema
- hand the agent real schemas instead of guesses
- assess the impact of a change before shipping it
Example prompts for the agent
- Find the orders table in the catalog and show me its owner
- Trace the lineage for the sales mart
- Which Kafka topics are cataloged and who owns them?
How to install Marmot Data Catalog
1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
claude mcp add --transport http marmot-data-catalog https://{marmot_host}/api/v1/mcp1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
codex mcp add marmot-data-catalog --url https://{marmot_host}/api/v1/mcp1Template address
The server address is a template: replace {…} or […] with your own values
2Config file
{
"mcpServers": {
"marmot-data-catalog": {
"url": "https://{marmot_host}/api/v1/mcp"
}
}
}1Template address
The server address is a template: replace {…} or […] with your own values
2Config file
{
"servers": {
"marmot-data-catalog": {
"type": "http",
"url": "https://{marmot_host}/api/v1/mcp"
}
}
}1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
gemini mcp add -t http marmot-data-catalog https://{marmot_host}/api/v1/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Marmot Data Catalog
How do I install Marmot Data Catalog in Claude Code?
Run claude mcp add --transport http marmot-data-catalog https://{marmot_host}/api/v1/mcp, replacing {marmot_host} with the address of your Marmot instance.
What does Marmot Data Catalog do?
It connects your agent to the Marmot catalog over MCP: asset search, lineage, ownership and metadata for tables, topics, queues and APIs.
What do I need for the server to work?
A deployed Marmot instance with the /api/v1/mcp endpoint reachable over HTTP. Without a running Marmot there is nothing to connect to.
