What problem does it solve? Writing long-form articles from model memory produces unsourced claims, generic AI-sounding prose, and weak SEO. This Skill grounds every claim in cited web research, enforces anti-slop vocabulary rules, and delivers publication-ready articles with schema and metadata. ## Core Features & Use Cases - STORM-inspired research pipeline: Parallel agents extract cited facts, generate reader personas, and analyze competitor gaps before any drafting begins. - Anti-slop and adversarial review: An independent reviewer agent checks hard/soft banned vocabulary, burstiness, and unsourced claims, followed by cross-model adversarial review. - SEO and site-aware output: Generates keyword-optimized meta tags, BlogPosting/FAQ JSON-LD schema, internal link suggestions, and frontmatter auto-detected from an existing site content directory. - Use Case: Run it with a topic, --site-dir pointing at your blog, and --keyword to produce a voice-matched, research-cited article saved directly into your content collection with matching frontmatter. ## Quick Start Ask the AI to write a 1500-word casual-tone article on your topic with the write-article skill, optionally passing --site-dir, --keyword, and --lang flags.
