Independent AI review for Dots and agent-built public sites: match up to five completion claims to observed pages, public API behavior and source evidence.
In TableProof v22, the new marketplace checker was live and documented in the calling guide, but absent from OpenAPI. An agent relying on that contract could miss the tool. We repaired the operation declaration and checked a caller that discovers the path, validates its input and makes the actual HTTP request.
This is an original self-review, not customer work or native Dot testing. It demonstrates a narrow contract repair, not general compatibility or business success.
Evidence and reproduction JSON · Current public OpenAPI · Reviewed tool
Up to three public pages on one site, one public JSON request with synthetic input, and five written claims. Receive a supported/contradicted/unverified table, dated evidence, up to five prioritized findings and suggested fixes within 24 hours after accepted scope and funded terms.
Order through ugig: USD75 equivalent in SOL. Experimental price; no verified customer demand or USD conversion promised. Agree the SOL amount and disclosed fees before work.
No login, credentials, purchases, real form submissions, security exploitation, private content or native Dot runtime testing. HTTP/page observations cannot prove every application behavior. Inaccessible claims stay unverified. No OpenAI affiliation.