# google-knowledge-graph [Health: Active]

**Category:** 🔎 Search & Data Extraction  
**Repository:** https://github.com/houtini-ai/google-knowledge-graph-mcp  
**GitHub Stars:** 13  
**npm Downloads (last month):** 1086  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/google-knowledge-graph

## Description
Search Google's Knowledge Graph for structured information about real-world entities

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "google-knowledge-graph": {
    "command": "npx",
    "args": ["-y","@houtini/google-knowledge-graph-mcp"],
    "env": {
      "GOOGLE_KNOWLEDGE_GRAPH_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `GOOGLE_KNOWLEDGE_GRAPH_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What google-knowledge-graph does

google-knowledge-graph exposes Google's Knowledge Graph Search API through MCP. It lets an AI client find structured records for real-world entities such as people, companies, locations, events, creative works, products, programming languages, sports teams, countries, and cities.

Search results can include an entity's name, schema.org-style types, short and detailed descriptions, image and web URLs, Google Machine IDs, and relevance scores. The returned data is JSON that an MCP client can parse for research or downstream entity-resolution tasks.

Two access patterns are available:

- Search using a name or descriptive query.
- Retrieve known entities using one or more Machine IDs.

## How it works

The server sends requests to Google's `https://kgsearch.googleapis.com/v1/entities:search` endpoint and passes the configured API key as a query parameter. Responses use JSON-LD and contain an `itemListElement` result array, which the server exposes as structured JSON.

Query searches accept an optional language list, entity-type filters, and a result limit. The default limit is 20 and the maximum is 500. Machine ID lookups accept an array of IDs and can also specify languages. Common type filters include `Person`, `Organization`, `Place`, `Event`, `CreativeWork`, `Product`, and `ComputerLanguage`; multiple types can be combined.

The server uses CommonJS output and is intended to run as a Node.js MCP process over the configuration supported by an MCP host.

## Setup and configuration

Create a Google Cloud project, enable the Knowledge Graph Search API, and create an API key under Credentials. Restricting the key to that API is recommended. The API key must be supplied through `GOOGLE_KNOWLEDGE_GRAPH_API_KEY`.

The simplest installation uses the npm package through `npx`:

```json
{
  "mcpServers": {
    "google-knowledge-graph": {
      "command": "npx",
      "args": ["-y", "@houtini/google-knowledge-graph-mcp"],
      "env": {
        "GOOGLE_KNOWLEDGE_GRAPH_API_KEY": "your-api-key-here"
      }
    }
  }
}
```

Claude Code can register the same command with `claude mcp add`, passing the API key as an environment option. Alternatively, clone the repository, run `npm install`, build with `npm run build`, and point the MCP configuration at `dist/index.js` using Node.js. Claude Desktop configuration files are stored under its platform-specific application-support location; Windows paths require escaped backslashes.

## Tools and capabilities

### `search_knowledge_graph`

Searches by a required query string. Optional `languages`, `types`, and `limit` parameters control language selection, entity filtering, and the number of results.

### `lookup_knowledge_graph_entities`

Resolves required Machine IDs supplied as an array. An optional `languages` parameter controls language selection for the returned records.

These tools support workflows that need ranked candidates, entity metadata, descriptions, official URLs, images, or IDs for later lookups.

## Limitations and notes

Requests are subject to Google's free-tier quotas; the documented general limit is approximately 100,000 queries per day. An API key is required, and a Google Cloud key must be used rather than a Google AI Studio key. The Knowledge Graph Search API must also be enabled in the associated project.

If searches return no results, broader query terms, fewer type filters, or a higher result limit may help. When the server is missing from Claude Desktop, check JSON syntax, environment-variable spelling, path escaping, and restart the application completely after changing configuration.

_Full upstream README: https://allmcps.com/mcp/google-knowledge-graph/readme_