Build the context layer your engineers and AI agents need

Guides and comparisons on cataloguing, lineage, governance and quality, and on giving AI agents governed context and governed access to data. Written by the team behind Marmot, the open source data catalog, query gateway and MCP server for agents and humans.

Compare data catalogs for AIGet started with Marmot

Browse by topic

• Data catalogWhat a data catalog is, why it matters, and how to choose one in 2026.
• Open source data catalogThe self-hosted options compared on licence, footprint, lineage and agent support.
• Data lineageTable and column-level lineage, OpenLineage and how lineage is collected.
• AI context layerWhat an AI context layer is, why agents need one, and how it differs from RAG.
• Data governanceOwnership, policies and access control for data and the agents that use it.
• AI agent database accessSQL for agents without database passwords: identity, read-only checks and a query audit.
• Data qualityTrust, freshness and certification, and why agents need it more than humans.
• AI data engineeringGiving LLMs and agents real context about your data landscape.
• MCP for dataThe Model Context Protocol, and how catalogs expose context to AI tools.

Latest

• AI Agent Database Access: Governed SQL Without PasswordsHow to let agents query your databases safely: their own identity, read-only SQL checked before it runs, least-privilege grants and an audit of every query.
• What Is Data Lineage? A 2026 Guide to Column-Level LineageTable versus column-level lineage, how lineage is collected with OpenLineage and lineage APIs, the tools that track it and why agents need it.
• Best Open Source Data Catalogs in 2026: Self-Hosted OptionsMarmot, OpenMetadata, DataHub, Apache Atlas, Marquez and the archived Amundsen compared on licence, footprint, lineage and AI agent support.
• Data Catalogs as the AI Context Layer: A 2026 ComparisonAn honest, side-by-side look at how Marmot, OpenMetadata, DataHub, Atlan, Collibra, Secoda, Amundsen and Apache Atlas expose governed context to AI agents over MCP and API.
• Marmot vs DataHubNative built-in MCP on a single binary versus DataHub's official MCP server on a Kafka, graph and search stack. Where each one wins, stated plainly.
• Marmot vs OpenMetadataTwo open source catalogs with native MCP, compared on footprint and breadth. A single Go binary on Postgres versus the widest connector platform on a search cluster and ingestion framework.
• Marmot vs AtlanAn open source catalog you self-host as a single Go binary versus Atlan's fully managed enterprise SaaS. Where each wins on MCP, cost, control and compliance.
• MCP for Data: Connecting AI Agents to Your CatalogWhat the Model Context Protocol is, how a catalog exposes context to tools like Claude and Cursor, and when to reach for MCP versus a plain API.
• AI Data Engineering: Giving Agents Real ContextWhy the model is rarely the limit, and what context an agent actually needs, schemas, ownership, lineage and glossary, to stop guessing about your data.