A bounded qualification of PrOntoQA-OOD: Finite generator capacity and citation-local scoring

This technical report presents a bounded qualification of one pinned PrOntoQA-OOD benchmark-generator configuration and a citation-local symbolic reasoning scorer. Two identical recorded passes each performed 1,574 generation attempts and returned 44 valid examples. Configurations requiring relevant distractors at depths 5 and 7 returned no examples. Source-level analysis establishes that accepted depth-7 examples are impossible under the exact tested configuration: at least 15 property families would be required, while only 13 are available. The scoring qualification includes 25 registered fixtures, a recorded aggregate of 5,120 logical cross-checks with zero reported mismatches, and 44 matched valid/invalid reasoning pairs. Individual cross-check comparison records were not retained. These findings are specific to the tested implementation and configuration. No language-model reasoning performance or recursive self-improvement was measured. The work does not claim established novelty or independent human peer review. The accompanying evidence archive contains the frozen experimental records, unsuccessful generation attempts, source snapshots, validation tools, reproducibility instructions, licensing information, and explicit AI-assistance disclosure. Research repository: https://github.com/seyedalimirmostafa-bit/prontoqa-q01-qualification