Best for
- Data teams with an OpenMetadata deployment
- AI assistants and agents that need governed data context
- Organizations performing discovery across semantics, lineage, quality, policies, and business terms
O
OpenMetadata
Official
Governed data context and business semantics for AI assistants and agents.
- Category
- Other #142 of 230
- Stars
- ★ 15.5k Very popular
- Transport
- Streamable HTTP
- Runtime
- Hosted (no local runtime)
- Credentials
- API key / credential required
- License
- Apache-2.0
- Last commit
51FMRS · D
Official OpenMetadata MCP server for AI assistants and agents that need governed data context, business semantics, lineage, quality, policy, and organizational memory from an OpenMetadata deployment. The source provides no specific tool identifiers or client support levels, so those lists remain empty.
Strongest · Maintenance 13/20 Weakest · Reliability 7/20
Why each score
- Reliability 7/20
- The manifest declares a configurable streamable-HTTP MCP endpoint and describes authentication, so the intended initialization path is plausible. However, the supplied material contains no server-specific tool list, init/tool-list handshake evidence, committed tests, CI workflows, or implementation-level error handling. Whether the declared behavior works cannot be verified, so the score is conservative.
- Security and permissions 10/20
- The manifest states that the endpoint uses OAuth 2.0 with PKCE or Basic Auth, and the README mentions roles, policies, and governed metadata actions, showing basic access-control intent. There is no evidence of token storage, permission scopes, tenant or data boundaries, sensitive-data flows, confirmation for dangerous writes, or audit behavior. This supports only a middling score.
- Maintenance 13/20
- The repository is not archived and the material indicates version 1.1.0, an Apache-2.0 license, ongoing commit activity, official documentation, and a clear repository owner. These are positive maintenance signals, but there is no concrete release history, dependency-update evidence, vulnerability-response process, or issue-response data, so full marks are not justified.
- Documentation 11/20
- The README gives a broad overview of capabilities, MCP use cases, authentication, endpoint path, and official documentation links, with examples of possible operations. It does not provide per-tool parameters, return formats, limits, failure diagnosis, cost information, or reviewable server-level examples. The README and manifest are also low-trust evidence under the rubric, so points are deducted.
- Setup experience 10/20
- The manifest provides an HTTPS streamable-HTTP URL with a host variable, default-port examples, the /mcp path, and OAuth/Basic Auth options, making the connection prerequisites reasonably clear. It lacks mainstream-client configuration examples, deployment prerequisites, platform compatibility details, and verifiable end-to-end steps from installation to a working connection, so setup remains manual.
Static review · not runListed 2026-07-29
Read the FMRS scoring method →Fit and risk
What it can accessUses the networkConnects to a database
Not for
- Users without an OpenMetadata deployment
- Use cases that only require direct access to a raw database or files
- Browser automation, cloud storage, or general collaboration workflows
Required permissions
- Access to the target OpenMetadata deployment's /mcp endpoint is required
- Authentication is required through OAuth 2.0 with PKCE and the existing SSO provider, or through Basic Auth
- The available metadata and data-quality operations are controlled by OpenMetadata authentication, authorization, roles, and policies
Risks and side effects
- The connection may expose permission-controlled metadata, lineage, quality, policy, and memory content from the deployment
- Accounts with write permissions may update descriptions, tags, owners, and other metadata, or create glossary terms, lineage, and quality tests
- Semantic-search results or memories may contain internal organizational knowledge and should be governed accordingly
Setup
Before you start
Runtime：Hosted (no local runtime)
Replace openmetadata_host with the hostname of an OpenMetadata deployment; include a non-standard port when needed, such as metadata.example.com:8585. Connect to https://{openmetadata_host}/mcp and authenticate through the existing OpenMetadata SSO provider using OAuth 2.0 with PKCE, or use Basic Auth.
Check that it works
Connect your MCP client to https://{openmetadata_host}/mcp with your OpenMetadata credentials and confirm the OpenMetadata MCP tools (e.g. search_metadata) appear in the client's tool list; then call search_metadata with a known asset name to prove the connection.
Troubleshooting
- Verify the openmetadata_host hostname and required port
- Confirm that the deployment is reachable at /mcp
- Check OAuth 2.0 with PKCE or Basic Auth configuration against OpenMetadata SSO or account settings
- Verify that the account has the required OpenMetadata roles and policy permissions
- Confirm that the client supports streamable HTTP
Things to try
Once connected, you can ask your AI assistant things like:
- Search metadata for customer datasets and show who owns them.
- Inspect the column-level lineage of this table and tell me what breaks downstream if it changes.
- What data contract and policies apply to this dataset, and is it fresh and certified?
- Create a data quality test for this table and save a memory note about the root cause of the last failure.
Use cases
Search data assets and metadata with natural language
Retrieve entity details, lineage, ownership, quality, and governance context
Retrieve or preserve organizational memories attached to data context
Support data-quality test creation, failure analysis, and root-cause analysis
Update descriptions, tags, and owners, or create glossary terms and lineage
Overview
The official OpenMetadata MCP server exposes an OpenMetadata deployment's metadata graph to MCP-compatible assistants and agents. Supported capabilities described by the source include metadata search, semantic search, entity details, lineage, data-contract and policy context, memory retrieval and preservation, metadata updates, glossary and lineage creation, data-quality tests, and root-cause analysis. This profile covers only openmetadata-mcp/server.json and does not attribute sibling repository servers to it. The server is version 1.1.0 and uses the deployment's /mcp endpoint.
Similar servers
Hevy MCP
77 · B
Manage your Hevy workout data from AI assistants via MCP.
Firecrawl MCP Server
75 · B
Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents
BioMCP
75 · B
One binary. One grammar. Evidence from the biomedical sources you already trust.
Compartment
75 · B
Encrypted, fully offline long-term memory for AI agents
More servers from this repository
Source revision 1fe0f038adb7 Data synced 2026-10-19 Read the FMRS scoring method