canonical: https://jentic.com/apis/googleapis.com/factchecktools # Google Fact Check Tools API The Fact Check Tools API exposes Google's index of ClaimReview-marked-up fact checks published across the open web. Developers can search published fact checks by claim text, by image, or by topic, and publishers can manage their own ClaimReview pages programmatically - creating, listing, updating, and deleting page entries that Google ingests into the index. The API is the canonical interface for both consuming third-party fact-check signals and contributing publisher claim reviews. ## For AI agents Search Google's index of fact checks by claim text or image, and manage publisher ClaimReview pages so they appear in Google's fact-check surfaces. ## Scope Does not assess claim truthfulness directly, host original articles, or moderate user-generated content - use for searching ClaimReview-marked fact checks and managing publisher pages only. ## Capabilities - Search published fact checks by free-text claim and language to find existing reviews from Google's indexed publishers - Search fact checks for an image by URL to surface visual claim reviews - Create a publisher ClaimReview page with one or more claim reviews so Google can index it - List the publisher's existing ClaimReview pages with paging and filtering - Update or delete a publisher ClaimReview page after the underlying article changes - Get a single ClaimReview page by name to inspect its content and review status ## Use cases ### Newsroom Claim Triage Newsroom researchers and fact-checking teams query the Fact Check Tools API to see whether a circulating claim has already been reviewed by a peer organisation before they spend time on a fresh check. The claims:search endpoint returns ClaimReview metadata including the rating, the review URL, and the publisher, letting researchers cite the existing review or differentiate their angle. This shaves hours off triage during fast-moving stories. Example prompt: Call GET /v1alpha1/claims:search with query='5g causes covid' and languageCode='en' and return the top three publishers, ratings, and URLs. ### Visual Misinformation Detection Trust and safety teams reviewing image-based misinformation use claims:imageSearch to find published fact checks tied to specific images circulating online. The API returns ClaimReviews that reference the image, including the review's textual rating, helping moderation pipelines surface authoritative context to users. Example prompt: Call GET /v1alpha1/claims:imageSearch with imageUri set to the questionable image URL and surface any matching ClaimReview entries to the moderator. ### Publisher ClaimReview Distribution Fact-checking publishers publish ClaimReview-tagged articles and use the API to register, list, and maintain those pages so Google's fact-check surfaces (Search, News, YouTube info panels) ingest them quickly and reliably. The same flow keeps stale or retracted pages out of the index when articles are taken down. Example prompt: POST /v1alpha1/pages with the article URL, publishing date, and claimReviewMarkups list, then list and verify the page appears in the publisher's pages collection. ### Agent-Mediated Claim Verification via Jentic An AI agent answering a user question that depends on a contested claim can call the Fact Check Tools API through Jentic to surface authoritative existing reviews before responding. The agent does not need to learn the claims:search or claims:imageSearch endpoint shapes; Jentic returns the matching operation with its query parameters per intent. Example prompt: Search Jentic for 'check if a claim has been fact-checked', load the GET /v1alpha1/claims:search schema, and execute it with the claim text before composing the answer. ## Key endpoints | Method | Path | Description | | --- | --- | --- | | GET | /v1alpha1/claims:search | Search fact checks by claim text | | GET | /v1alpha1/claims:imageSearch | Search fact checks by image URL | | GET | /v1alpha1/pages | List publisher ClaimReview pages | | POST | /v1alpha1/pages | Create a ClaimReview page | | GET | /v1alpha1/{+name} | Get a single ClaimReview page | | PUT | /v1alpha1/{+name} | Update a ClaimReview page | | DELETE | /v1alpha1/{+name} | Delete a ClaimReview page | ## Key resources - **Claims** — Search published fact checks by text query or image URL - **Pages** — Create, list, get, update, and delete publisher ClaimReview pages ## Why Jentic - **Setup:** Wiring the Fact Check Tools API by hand means standing up Google OAuth2 with the right scopes, targeting factchecktools.googleapis.com, and building the claim search and page calls yourself. Through Jentic you install once, import the Fact Check Tools API from the API Directory, store the OAuth credential once, and your agent calls it. - **Permission scoping:** Fact Check Tools puts the page name in the URL path (/v1alpha1/{+name}) while claim searches take their query in the request, so a rule can pin your agent to one publisher page for reads and updates, and you choose the operations it may call so creating or replacing pages is not included unless you add it. Its read operations only cover the pages and searches you allow. - **Credential handling:** Your Google OAuth credential for Fact Check Tools is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'search ClaimReview fact checks' or 'list publisher fact-check pages', and Jentic returns the matching Fact Check Tools operation with its input schema so the agent calls the right endpoint without browsing the reference docs. ## Related APIs - **Custom Search API** — Web search to find articles around a claim before checking against the fact-check index - **Search Console API** — Verifies publisher domain ownership for ClaimReview submission - **Books API** — Authoritative reference data for verifying claims about published books ## FAQ ### What authentication does the Fact Check Tools API use? The Fact Check Tools API uses Google OAuth 2.0 with the userinfo.email scope. Search endpoints accept user OAuth tokens or API keys depending on quota; publisher endpoints require an OAuth token issued to a verified publisher account. Through Jentic the credentials live in the encrypted vault. ### Can I search for fact checks about a specific image? Yes. GET /v1alpha1/claims:imageSearch accepts an imageUri query parameter and returns ClaimReview entries that reference that image, including the review URL and rating. This is useful for visual misinformation triage. ### What are the rate limits for the Fact Check Tools API? The API enforces standard Google per-project quotas in the Cloud Console (typical default 600 read requests per minute). Search results are bounded per call and pageToken-paginated; publisher endpoints are scoped to the publisher's verified domain. ### How do I publish a ClaimReview page through Jentic? Search Jentic for 'publish a ClaimReview page', load the POST /v1alpha1/pages schema, and execute it with the page URL and claimReviewMarkups list. The publisher account must be verified in the Fact Check Tools console first. ### Is the Fact Check Tools API free? Yes. The API is free for both search and publisher operations within standard quota limits. There is no per-call charge for either consuming fact checks or registering publisher ClaimReview pages. ### How do I find fact checks in a specific language? Pass the languageCode query parameter to GET /v1alpha1/claims:search (e.g. 'en', 'es', 'pt') to restrict results to ClaimReviews in that language. Combine with reviewPublisherSiteFilter to scope results to specific publishers. ### Can I limit what my agent is allowed to do with the Fact Check Tools API? Yes. Because you run Jentic One yourself, your own rules decide which Fact Check Tools operations and OAuth credential the agent may use. You can allow just the read operations, such as GET /v1alpha1/claims:search and GET /v1alpha1/claims:imageSearch, while excluding the create, update, and delete operations on /v1alpha1/pages unless you add them. Since a page is addressed by name in the URL path (/v1alpha1/{+name}), a rule can also pin the agent to a single publisher page for reads and updates.