What problem does it solve? AI-generated drafts carry recognizable tells — em dashes, rule-of-three structures, promotional adjectives, vague attributions — that make public-facing content read as machine-written and erode audience trust. This Skill scans text against 24 documented patterns from Wikipedia's "Signs of AI writing" guide and rewrites it to sound authentically human while preserving meaning. ## Core Features & Use Cases - 24-pattern detection: Scans for AI vocabulary, copula avoidance, negative parallelisms, inline-header bold lists, title-case headings, curly quotes, emojis, sycophantic phrases, hedging, and filler. - Voice injection: Goes beyond removal — adds opinions, varied sentence rhythm, first-person perspective, and specific details so the text has personality rather than sterile neutrality. - Brand alignment: Cross-checks output against Context/Brand.md for organisation-specific tone, language standard, and sector rules. - Use Case: Before publishing a LinkedIn post or client proposal drafted with AI assistance, run this pass to strip AI cadence, then receive the rewritten text plus a summary of which patterns were found and fixed. ## Quick Start Humanise this draft newsletter copy and remove any signs of AI-generated writing before I publish it.