canonical: https://jentic.com/apis/googleapis.com/kgsearch

# Google Knowledge Graph Search API

The Knowledge Graph Search API exposes a single endpoint that searches Google's Knowledge Graph for entities - people, places, organizations, films, and other notable things - and returns Schema.org-formatted results with names, descriptions, types, image URLs, and detailed descriptions. It is intended for autocomplete experiences, entity disambiguation, and enrichment of internal records with canonical Google identifiers. The API is read-only and works with a simple API key.

## For AI agents

Search the Google Knowledge Graph for entities and return Schema.org JSON-LD results with names, types, descriptions, and images. Useful for entity enrichment and autocomplete.

## Scope

Does not write to the Knowledge Graph, return web search results, or expose entity relationships beyond the @type field - use only for read-only entity lookup.

## Capabilities

- Search the Knowledge Graph by free text query and return ranked entity matches
- Filter entity results by Schema.org type such as Person, Organization, or Movie
- Restrict results to a specific language for localized names and descriptions
- Limit results to a maximum number of entities per query
- Look up an entity directly by its Knowledge Graph machine ID
- Receive Schema.org JSON-LD output with image URLs and detailed descriptions

## Use cases

### Entity enrichment for CRM records

Marketing and sales operations teams enrich company and contact records with stable Knowledge Graph IDs so multiple internal systems can refer to the same canonical entity. A single GET /v1/entities:search call with a company name and types=Organization returns ranked candidates with descriptions and image URLs the agent can use to confirm a match.

Example prompt: Search the Knowledge Graph for query 'Stripe' with types=Organization and limit=3, then return the top result's @id, name, and description.

### Autocomplete for entity pickers

Apps that let users tag content with people, films, or places call entities:search as the user types. Each request returns ranked entity matches with image URLs and short descriptions, which is enough to render a rich dropdown without maintaining an internal entity database.

Example prompt: On each keystroke, call entities:search with the partial query and limit=5, then render the resulting names and image URLs in the dropdown.

### Disambiguating ambiguous names

When a user types a query like 'Mercury' the same string can refer to a planet, a chemical element, a band, or a brand. The Knowledge Graph Search API returns ranked candidates with type and description so an agent can surface a disambiguation list rather than guessing which Mercury the user meant.

Example prompt: Call entities:search?query=Mercury&limit=5 and present each result's @type and description so the user can pick.

### Agent-driven knowledge enrichment via Jentic

An agent answering questions about real-world entities can call the Knowledge Graph Search API through Jentic without managing API keys. Jentic surfaces the single search operation in natural language and returns the schema, so the agent can cite Google's canonical entity description when grounding its answer.

Example prompt: When asked 'who is Hedy Lamarr', call google_kgsearch_query with query='Hedy Lamarr' and return the top result's detailedDescription.articleBody.

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| GET | /v1/entities:search | Search the Knowledge Graph for entities by query, types, language, and limit |

## Key resources

- **entities** — Search the Knowledge Graph by query text or machine id and return Schema.org-formatted results

## Why Jentic

- **Setup:** Wiring the Knowledge Graph Search API by hand means provisioning a Google Cloud API key, appending it as the key query parameter on every call, and shaping its entity-search query yourself. Through Jentic you install once, import the Knowledge Graph Search API from the API Directory, store the API key once, and your agent calls it.
- **Permission scoping:** This API exposes a single read-only entity-search operation and passes its terms as query parameters rather than a resource in the URL path, so scope the agent to that operation. Because the only endpoint reads entity data, the agent can look up entities but can never write to the Knowledge Graph.
- **Credential handling:** Your Google Cloud API key is stored once, encrypted, by your own Jentic One instance and injected at execution time as the key query parameter. The raw key value never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'search the google knowledge graph' or 'look up an entity by name', and Jentic returns the entities:search operation with its query, types, and limit parameters so the agent calls it correctly without browsing the reference docs.

## Related APIs

- **Custom Search API** — Custom Search returns web search results; Knowledge Graph Search returns canonical entity records
- **Cloud Natural Language API** — Natural Language extracts entities from text; Knowledge Graph Search resolves names to canonical IDs
- **Clearbit API** — Clearbit returns commercial firmographic data on companies and people; Knowledge Graph Search returns Google-curated public entity records

## FAQ

### What authentication does the Knowledge Graph Search API use?

A simple API key passed as the key query parameter. The spec lists no OAuth scheme, so the only credential is a Google Cloud API key with the Knowledge Graph Search API enabled. Through Jentic the API key sits in the encrypted vault and is injected at execution time.

### What types can I filter Knowledge Graph results by?

The types parameter accepts any Schema.org type, with Person, Organization, Place, Movie, Book, and Event being the most common. Pass multiple types to broaden the search, e.g. types=Person&types=Organization.

### What are the rate limits for the Knowledge Graph Search API?

Google's default quota is 100,000 queries per day per project with a per-second burst limit set per API key. Quotas can be raised in the Cloud console for high-volume autocomplete use cases.

### How do I look up an entity through Jentic?

Run jentic search 'search the google knowledge graph' to find GET /v1/entities:search, jentic load to fetch its parameters, then jentic execute with query, types, languages, and limit. Jentic handles the API key transparently.

### Can I retrieve an entity by its Knowledge Graph ID?

Yes. Pass the machine id as the ids query parameter on /v1/entities:search to fetch a specific entity directly without a free-text query.

### Is the Knowledge Graph Search API free?

There is no per-call charge - usage is bounded only by the daily quota of 100,000 queries per project. Higher tiers require a quota increase request through Google Cloud.

### Can I limit what my agent is allowed to do with the Knowledge Graph Search API?

Yes. This API exposes a single read-only operation, the GET /v1/entities:search entity lookup, and because you run Jentic One yourself your own rules decide which operations and credentials your agent may use. You can scope the agent to just that search operation, so it can look up entities by query, types, language, or machine id but can never write to the Knowledge Graph. Your Google Cloud API key is stored once by your own instance and injected at execution time, so the agent calls the endpoint without ever seeing the raw key.