hallucination-detector

The hallucination-detector skill presents a coherent and proportionate footprint for its stated purpose. It primarily loads a single source of truth (tech-stack.md), validates all technical suggestions, requires evidence, and blocks/corrects hallucinations. The data flows, dependencies, and potential risk signals are aligned with its goal of preventing misaligned technology decisions. Some edge cases (doc drift, registry outages, generated schemas) warrant explicit handling, but none imply fundamental misalignment with the stated objective. Overall, the design is Benign with low to moderate security risk, and it should function as intended to reduce wasted effort and ensure architectural consistency.