概要
This skill equips Claude with a sophisticated, multi-layer methodology for identifying AI-generated content based on academic research and empirical analysis. It scans for high-signal vocabulary, structural patterns like uniform sentence lengths (low burstiness), and technical artifacts specific to models such as GPT-4, Gemini, and Grok. Designed for high-stakes environments, the skill emphasizes false-positive prevention—particularly for non-native English speakers—and provides structured reports that categorize evidence into technical, vocabulary, and formatting signals to ensure objective assessments.
