The data catalog that ships as one binary and feeds your AI

What it does

Marmot is an open-source data catalog written in Go that indexes tables, topics, queues, APIs, and dashboards. It bundles full-text search, interactive lineage tracing, and team collaboration into a single binary with a web UI, aiming to cut the deployment time from days to minutes. You catalog assets, enrich them with context, and make the result accessible to your team.

The interesting bit

Instead of treating a catalog as a passive search portal for humans, Marmot positions itself as a context layer for AI agents. It exposes certified metadata through an MCP interface—plus a REST API and UI—so AI agents can consume catalog context alongside human users.

Key highlights

• Ships as a single binary with an embedded UI, sidestepping the usual distributed deployment headache.
• Traces data lineage interactively and analyses impact before you break a downstream dashboard.
• Supports full-text search with boolean logic and metadata filters across every asset type.
• Exposes catalog context to AI agents via an MCP interface alongside its REST API and UI.
• Lets teams assign ownership, document business context, and maintain shared glossaries.

Verdict

Data engineers and platform teams tired of wrestling with enterprise catalog sprawl should look here; if you need a fully managed SaaS with SLAs or deep governance workflows, this is still a self-hosted open-source project.