Wikidata MCP

The Wikidata MCP (Model Context Protocol) provides a set of standardized tools that allow large language models (LLMs) to explore and query Wikidata programmatically. It is designed for agentic AI or AI workflows that need to search, inspect, and query Wikidata, without relying on hardcoded assumptions about its structure or content.

The Wikidata MCP server is running at https://wd-mcp.wmcloud.org/
You can connect your AI application to it at https://wd-mcp.wmcloud.org/mcp
Tools are exposed as API endpoints and can be tested interactively at https://wd-mcp.wmcloud.org/docs

🧰 Tools

1. search_items(query: str, lang: str = "en") -> str
   Searches Wikidata items (QIDs) using vector search when available and falls back to keyword search when needed. Returns matching QIDs with labels and descriptions.

Use When: Starting exploration from a concept or natural-language description.

1. search_properties(query: str, lang: str = "en") -> str
   Searches Wikidata properties (PIDs) using vector search when available and falls back to keyword search when needed. Returns matching PIDs with labels and descriptions.

Use When: You need to find the right Wikidata property for relationships in statements or SPARQL.

1. get_statements(entity_id: str, include_external_ids: bool = False, lang: str = "en") -> str
   Returns direct statements (property-value pairs) for an entity in triplet-like text form. This tool excludes qualifiers, references, and deprecated values.

Use When: You want a fast structural overview of an entity.

1. get_statement_values(entity_id: str, property_id: str, lang: str = "en") -> str
   Returns all statement values for an entity-property pair, including qualifiers, references, and all ranks.

Use When: You need full statement detail for auditing, fact-checking, or provenance-sensitive tasks.

1. get_instance_and_subclass_hierarchy(entity_id: str, max_depth: int = 5, lang: str = "en") -> str
   Retrieves hierarchical context using "instance of" (P31) and "subclass of" (P279), returning JSON-formatted hierarchy data.

Use When: You need to understand entity classification before building filters in SPARQL.

1. execute_sparql(sparql: str, K: int = 10) -> str
   Executes a SPARQL query against Wikidata and returns up to K rows as CSV text.

Use When: You want structured retrieval and verification from Wikidata Query Service.

🚀 Running Locally

Run:

uv run python main.py

Then open:

• http://localhost:8000/ for project page
• http://localhost:8000/docs for interactive Swagger UI
• http://localhost:8000/mcp for MCP clients

With Docker:

docker compose up --build

🌐 Services

Vector Search

This service interfaces with the Wikidata Vector Database, enabling semantic search over Wikidata items using natural language. It is ideal for discovering relevant items without needing to know exact labels. This serves as a first step in exploratory or context-rich workflows.

🚀 API: wd-vectordb.wmcloud.org
📚 Docs: wd-vectordb.wmcloud.org/docs
📄 Project Page: Wikidata Embedding Project

Wikidata Textifier

This service returns readable triplet or textual representations of Wikidata entities, with resolved labels, optimized for use by language models.

🚀 API: wd-textify.wmcloud.org
📚 Docs: wd-textify.wmcloud.org/docs