Reasoning Under Load · 01: Claude Opus 4.8 — An Independent Reasoning-Integrity Evaluation

An independent evaluation of Claude Opus 4.8 applying seven content-neutral inference primitives as a reasoning-integrity rubric across a controlled five-condition experimental series. Three dimensions tested: reasoning-primitive integrity under dispositional load, memory-layer infrastructure behavior, and artifact coherence under extended collaboration. Case-study evaluation and framework proposal for LLM reasoning-integrity assessment beyond standard benchmarks.