MCP OpenMetadata

Name: MCP OpenMetadata
Author: pfldy2850

@pfldy2850

リポジトリ

MCP OpenMetadata について

mcp-openmetadata

#mcp-openmetadata #metadata-management #api-integration

基本情報

カテゴリ

その他

ライセンス

MIT

ランタイム

python

トランスポート

stdio

公開者

pfldy2850

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "mcp-openmetadata": {
      "command": "uv",
      "args": [
        "pip",
        "install",
        "mcp-openmetadata"
      ]
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is MCP OpenMetadata?

MCP OpenMetadata is a FastMCP-based server that provides a subset of OpenMetadata APIs for querying and managing metadata. It offers tools for searching entities, retrieving table metadata, columns, ownership information, and sample data—designed for AI assistants to interact with an OpenMetadata instance.

How to use MCP OpenMetadata?

Install the package via uv pip install mcp-openmetadata or pip install mcp-openmetadata. Configure it in your MCP client with the command python -m mcp_openmetadata.server and environment variables for authentication (OPENMETADATA_URI plus either OPENMETADATA_JWT_TOKEN for token auth or OPENMETADATA_USERNAME/OPENMETADATA_PASSWORD for basic auth). Then invoke any of the provided tools from your LLM.

Key features of MCP OpenMetadata

Search entities using query text with pagination and sorting
Retrieve paginated list of tables with basic information
Get detailed table information by fully qualified name
Fetch table columns and ownership details
Access sample data from specified tables
Supports token and basic authentication methods

Use cases of MCP OpenMetadata

An AI assistant listing all database tables in an OpenMetadata instance
Finding tables by column name or tag to locate relevant data assets
Retrieving column schemas and ownership for data governance checks
Getting sample rows to understand table content before querying
Integrating metadata discovery into automated data lineage or catalog workflows

FAQ from MCP OpenMetadata

How does it compare to using the OpenMetadata API directly?

The server returns optimized, context-friendly results instead of raw API responses that may contain excessive data. Not all OpenMetadata APIs are exposed; see the Supported APIs document for the list.

What are the runtime dependencies?

Python 3.10 or later, plus the fastmcp, httpx, and mcp-openmetadata packages.

Where does the server connect to?

It connects to an OpenMetadata instance via the configured OPENMETADATA_URI environment variable (default http://localhost:8585). No external cloud service is involved.

What authentication methods are supported?

Two options: token authentication (using OPENMETADATA_JWT_TOKEN) or basic authentication (using OPENMETADATA_USERNAME and OPENMETADATA_PASSWORD).