MLC Bakery

What it is

A service you run yourself with Docker (by default on your own computer) that catalogues ML datasets in collections, logs activity, and tracks where ML models came from and what they were built from. For each dataset it keeps a record: where the data lives, its Croissant metadata and, if one was uploaded, a preview; the data itself stays where it is. Its MCP server connects an AI assistant to that catalogue.

What you get

• Search across the datasets catalogued in your bakery
• A link to download a preview of a dataset
• A dataset's Croissant metadata, the metadata.json file that describes it
• A check of a Croissant metadata file: pass or fail, with the error when it fails
• Help on the MLC Bakery API and the server's tools, inside the chat

Requirements

Docker with docker compose. Copy env.example to a new file named .env, start the containers, then create the database. You may need to add mcp.localhost to your computer's /etc/hosts file so the assistant can reach the server. No outside account: the tools sign in with an admin token you set yourself in .env.

Setup effort

Run a container — build the image from the repository, then point your client at it

Requires

• Docker with docker compose
• A .env file copied from env.example
• An /etc/hosts entry for mcp.localhost

Where it runs

The documented client runs the server on your machine.