OpenMetadata

OpenMetadata is an open-source platform for data cataloging and metadata management, positioned by its maintainers as an open context layer for AI. It connects technical metadata, lineage, quality signals, ownership, glossaries, and organizational memory into one metadata knowledge graph. The project is Apache-2.0 licensed, and the most recent release listed is 2.0.5 (October 8, 2026). Collate offers a managed service.

What It Is

OpenMetadata ingests metadata from warehouses, lakes, BI tools, pipelines, ML platforms, messaging systems, and storage through 130+ connectors, and normalizes it with open, schema-first specifications. The result is a graph linking assets, columns, owners, teams, policies, quality tests, lineage, glossary terms, data contracts, data products, and memories. Users and AI agents can use it to discover, understand, and trust data.

Context, Semantics, and Memory

The site describes three primitives in the graph: context (schemas, tables, dashboards, pipelines, ML models, owners, lineage, classifications), ontology (glossary terms, metric definitions, and W3C-standard relationships such as RDF, OWL, DCAT), and memory (corrections, decisions, approvals, and an auditable change record). Memories are governed entities that can be attached to assets and captured through APIs, SDKs, MCP, or chat.

Agent and Developer Access

An MCP server lets compatible assistants search metadata, run semantic search, inspect lineage, update descriptions, tags, and owners, create glossary terms and lineage, and create data quality tests. The Python package openmetadata-ingestion reads and writes metadata, and the AI SDK (Python, TypeScript, Java) exposes MCP tools to agents, including as LangChain tools. REST APIs, events, and webhooks support automation.

Standards and Governance

OpenMetadata offers 700+ JSON Schema and RDF/JSON-LD specifications and supports OpenLineage and ODCS 3.1 data contracts. Governance features include classifications, policies, roles, certification, and review workflows. Data quality covers tests, profiling, freshness and volume checks, alerts, and incidents.

Update: 2.0 and 2.0.5

OpenMetadata 2.0 was announced as the open context layer for AI agents. The 2.0.5 maintenance release fixes connector and lineage issues (Redshift, Oracle, Mode, Databricks), keeps ingestion bots from overwriting user-managed tags and owners, and updates Jackson dependencies for two security vulnerabilities.

Deployment Options

You can deploy it yourself using the local Docker quickstart, try a live sandbox with demo data, or use the managed service from Collate.