elly99-AI/MarCognity-AI: Epistemic Stress Tests on Closed LLMs-Neuropsychological Perspective

This release includes a neuropsychology‑inspired protocol for evaluating the epistemic robustness of closed‑source LLMs. The test exposes the dissociation between linguistic fluency and epistemic grounding by analyzing model behavior under controlled academic tasks, revealing confabulations, unsupported claims, and the absence of metacognitive self‑monitoring. The protocol is model‑agnostic and fully reproducible.