OpenMetadata is an open-source metadata platform that connects technical metadata, data quality signals, lineage, ownership, usage, policies, glossaries, classifications, metrics, domains, data contracts, data products, and captured conversations into one metadata knowledge graph. It is positioned as a context layer for data users, AI assistants, and agents: a warehouse connection exposes raw structures, while OpenMetadata answers what a dataset means, who owns it, whether it is fresh, tested, and certified, which contract governs it, and what breaks downstream if it changes.
The platform is built around five capabilities: context (technical, operational, trust, usage, and lineage metadata), semantics (glossaries, metrics, classifications, domains, policies, ontologies, and data products), a knowledge graph of relationships between all of these, memory (conversations, AI threads, decisions, assumptions, runbooks, and remediation notes stored as governed memory entities attached to assets), and activation through Model Context Protocol (MCP), semantic search, APIs, SDKs, events, and workflows. Metadata is collected through 130+ connectors and ingestion APIs, normalised against open schemas, connected into the graph, governed with policies and review workflows, and then exposed to tools and agents.
For agent builders, the MCP server lets assistants search metadata semantically, retrieve entity details, inspect lineage, read data contracts and policy context, store or retrieve memories, update descriptions, tags, and owners, create glossary terms and lineage, and list, create, or root-cause data quality tests. Two Python packages cover code access: openmetadata-ingestion for reading and writing metadata, lineage, glossary, and quality, and data-ai-sdk for giving an LLM or agent governed access through MCP or LangChain.
Features
- Connectors: 130+ integrations across warehouses, lakes, BI tools, pipelines, ML platforms, messaging, storage, APIs, search, and SaaS applications
- Technical metadata: databases, schemas, tables, columns, topics, dashboards, charts, pipelines, APIs, search indexes, ML models, and storage assets
- Quality and observability: test cases and suites, freshness, volume, null, uniqueness, and distribution checks, profiling, incidents, and alerts
- Lineage: table, column-level, dashboard, pipeline, metric, and ML model lineage, including OpenLineage events
- Semantics: glossaries, business terms, synonyms, metrics, KPIs, classifications, tags, domains, and ontologies
- Governance: owners, stewards, roles, policies, certification, review workflows, lifecycle states, and data contracts
- Memories: governed memory entities tied to assets, users, teams, threads, domains, and workflows, retrievable via API, SDK, MCP, or chat
- MCP server: natural-language read and write access to the graph for MCP-compatible assistants
- Semantic search: search by meaning across assets, metrics, glossary terms, data products, and memories
- APIs, SDKs, and events: ingest, update, search, subscribe, and automate through REST APIs, SDKs, webhooks, and an AI SDK
- Open standards: DCAT, DPROD, PROV-O, OpenLineage, ODCS, RDF/OWL, JSON-LD, SHACL, and JSON Schema
