🔒All inputs are processed locally in your browser. Nothing is transmitted, stored, or logged. Inputs disappear when you close the tab.

OCG-Industries Showcase · Cluster ④ AI Agents and Memory

Claim Source Audit

A generated answer is a set of claims, and each one either names a retrieved source or it does not. Enter the sources that were retrieved and the claims that were generated with the source each cites, and this page finds every claim whose cited source was never retrieved. Those are the unsupported ones, and the page names them.

Data: no external data

Inputs STEP 1
One source id per line, exactly as the retriever returned them.
One per line as claim-id:source-id. Use claim-id:none for a claim that cites nothing.
Audit Status
--
Grounded only when every claim cites a retrieved source
Grounding
--%
Supported claims / total claims
Audit Key
--
Fingerprint of the answer, sources, and each claim's citation
Claims
No claims entered--
Detail
Answer reference--
Retrieved source count--
Claim count--
Supported / unsupported--
Uncited claims--
Dangling citations--
Unused retrieved sources--
Unsupported claim ids--
Audit key--
execution_hash: --
Grounding here means a claim cites a source that was retrieved. It does not mean the source supports the claim. Whether the cited text actually says what the claim says is outside what this page can check. Try changing one cited source id by a character and watch that claim fall out of the grounded set.
Policy Mandate Record v2.0 · what is this?