marmot

Why we included this project

Marmot is a data catalog that skips the usual infrastructure overhead: it ships as a single binary, so a team can have it running in minutes instead of standing up a stack of services. You register tables, topics, queues, APIs, and dashboards, then attach ownership, business context, and shared glossaries to each asset, which keeps the metadata useful to the people who actually work with it. What sets it apart is the MCP integration: the same curated context is exposed to AI agents, so an assistant can look up what a column means or follow a data flow rather than guessing. That combination of human-friendly discovery and agent-ready metadata makes it a good fit for data engineering and platform teams. The interactive lineage view is a nice bonus for checking the impact of a schema or pipeline change before you make it.