This paper presents a reproducible experiment in vibe-coding and vericoding. The authors used Anthropic’s Claude to solve the first 33 of the Ninety-Nine Prolog Problems from informal English specifications. The generated artifacts included 58 logic procedures, 508 tests, 257 lemmas, and 11,800 proof lines. They ran the tests, manually inspected the generated files and logical statements, and proof-checked the proofs with LPTP, the Logic Program Theorem Prover. The reported guarantees cover types, groundness, termination, uniqueness, existence, and, in some cases, functional correctness.
No heat snapshots are available in the last 24 hours.