Added the adversarial-challenge skill that red-teams a single blog post or page against its own front-matter and stated thesis, producing a calibrated Markdown findings report.
Give the author a harder, falsification-first probe of a draft than a re-read provides, for the increasingly argument-driven posts on the blog. Adapted from the AI Fundamentals Playbook’s chapter-adversary skill but retargeted from a book.yml contract to a single article’s own contract (front-matter title/description/categories plus the opening thesis), and stripped of all book/cross-chapter machinery. The skill reports only; it does not rewrite the prose, which is english-correction’s job. Executed as Step 2 of plan-002.
.claude/skills/adversarial-challenge/SKILL.md (new)execute plan 2