mcp-wikidata

The Wikidata MCP Server exposes tools to interact with Wikidata through the Model Context Protocol (MCP). It provides a set of tools to search for Wikidata identifiers, fetch metadata, and run SPARQL queries against Wikidata. Specifically, you can use: search_entity(query: str) to find entity IDs, search_property(query: str) to locate property IDs, get_properties(entity_id: str) to retrieve an entity's properties, execute_sparql(sparql_query: str) to run custom SPARQL queries, and get_metadata(entity_id: str, language: str = "en") to obtain the English label and description. Clients using MCP can chain these tools to, for example, identify an entity, inspect its properties to determine a suitable property for a query, and fetch results or metadata for presentation. The server supports practical workflows such as extracting a label/description in English for display and running SPARQL to fetch related entities or attributes.

Prerequisites:

• git
• Python environment (recommended) or uv runtime as per your setup
• Internet access to install dependencies from Wikidata and PyPI

Manual installation steps:

1. Clone the repository:

git clone https://github.com/zzaebok/mcp-wikidata.git
cd mcp-wikidata

2. Install runtime (example using uv) if not already installed:

# On macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

3. Install dependencies (Python project setup assumed by uv workflow):

uv sync

4. Run the server (as shown in the README example):

uv run src/server.py

If you want to test with a sample client:

uv run src/client.py

Alternative: If you prefer Smithery for automated deployment, install via Smithery:

npx -y @smithery/cli install @zzaebok/mcp-wikidata --client claude

Notes and tips:

• The server relies on Wikidata’ s public API for search, properties, and SPARQL execution. Ensure network access is allowed to wikidata.org.
• If you encounter language-specific metadata, try get_metadata with language set to the desired ISO language code (default is en).
• When testing SPARQL, keep queries simple to avoid long runtimes and to ensure reliable responses in client workflows.
• If using the uv-based run method, ensure uv is properly installed and available in your PATH. The README provides a quick install script for uv.
• The MCP tool list includes: search_entity, search_property, get_properties, execute_sparql, get_metadata. These can be composed in a client pipeline to perform end-to-end tasks such as finding a director, pulling their entity, and querying for related films.