google-knowledge-graph — Claude Code MCP server for entity search
google-knowledge-graph connects an agent to Google's Knowledge Graph API. Returns structured data about people, places and organisations during research tasks.
MIT ★ 13 updated 2026-10-10
claude mcp add google-knowledge-graph -e GOOGLE_KNOWLEDGE_GRAPH_API_KEY=<GOOGLE_KNOWLEDGE_GRAPH_API_KEY> -- npx -y @houtini/google-knowledge-graph-mcpWhat google-knowledge-graph does
The server exposes two tools. search_knowledge_graph finds entities by name or description, lookup_knowledge_graph_entities fetches a record by Machine ID. Results come back as JSON with names, schema.org types, Wikipedia links, images, relevance scores and MIDs for follow-up calls.
You run it through npx, no build step needed. The API itself is free, but the server will not start without a GOOGLE_KNOWLEDGE_GRAPH_API_KEY in the environment.
Type filtering helps when a query like “Python” matches several things. You can pass ["ComputerLanguage"] and narrow the result set. The limit parameter goes up to 500, languages can be pinned to a single code.
What it helps with
- verify entity facts while the agent works on research
- disambiguate names that match multiple real-world things
- pull structured descriptions and Wikipedia URLs into a workflow
- resolve an entity by Machine ID instead of searching again
Example prompts for the agent
- Search the knowledge graph for Marie Curie
- Find the entity Python with type ComputerLanguage
- Look up knowledge graph entity /m/0dl567
How to install google-knowledge-graph
Platform CLI
claude mcp add google-knowledge-graph -e GOOGLE_KNOWLEDGE_GRAPH_API_KEY=<GOOGLE_KNOWLEDGE_GRAPH_API_KEY> -- npx -y @houtini/google-knowledge-graph-mcpReplace <…> with your own values
Platform CLI
codex mcp add google-knowledge-graph --env GOOGLE_KNOWLEDGE_GRAPH_API_KEY=<GOOGLE_KNOWLEDGE_GRAPH_API_KEY> -- npx -y @houtini/google-knowledge-graph-mcpReplace <…> with your own values
Config file
{
"mcpServers": {
"google-knowledge-graph": {
"command": "npx",
"args": [
"-y",
"@houtini/google-knowledge-graph-mcp"
],
"env": {
"GOOGLE_KNOWLEDGE_GRAPH_API_KEY": "<GOOGLE_KNOWLEDGE_GRAPH_API_KEY>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"google-knowledge-graph": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@houtini/google-knowledge-graph-mcp"
],
"env": {
"GOOGLE_KNOWLEDGE_GRAPH_API_KEY": "<GOOGLE_KNOWLEDGE_GRAPH_API_KEY>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e GOOGLE_KNOWLEDGE_GRAPH_API_KEY=<GOOGLE_KNOWLEDGE_GRAPH_API_KEY> google-knowledge-graph npx -y @houtini/google-knowledge-graph-mcpReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about google-knowledge-graph
How do I install google-knowledge-graph in Claude Code?
Run claude mcp add google-knowledge-graph -e GOOGLE_KNOWLEDGE_GRAPH_API_KEY=<GOOGLE_KNOWLEDGE_GRAPH_API_KEY> -- npx -y @houtini/google-knowledge-graph-mcp. Check with claude mcp get google-knowledge-graph; the status should read Connected.
What does google-knowledge-graph do?
It gives an MCP client access to the free Google Knowledge Graph Search API. The server returns structured entity data: types, descriptions, images, Wikipedia links and Machine IDs.
Do I need a paid Google Cloud account?
No. Create a project in Google Cloud Console, enable the Knowledge Graph Search API and generate a key. Billing and card details are not required.