What problem does it solve? Producing consistent, well-researched technical articles on a schedule is hard: topics get repeated, drafts lack evidence, and tone drifts into marketing filler. This Skill automates the full editorial pipeline β topic selection, research, falsifiable thesis, drafting, self-editing, logging, and notification β so every article ships with citations and a defensible claim. ## Core Features & Use Cases - Three article angles: a standard long-form article or technical explainer on a trending or given topic, a thesis-driven article about a watched GitHub repo using gh api activity data, and a project-through-a-lens essay connecting the project to an external trend or idea. - Editorial quality gates: falsifiable thesis requirements, banned-phrase lexicons, source minimums (primary sources, in-repo and external links), word-count bounds, and abort-on-weak-draft logic. - Optional hero images: append --visual to generate a labeled technical illustration via Replicate's Nano Banana Pro, with a Seedream fallback and graceful text-only degradation when REPLICATE_API_TOKEN is absent. - Use Case: Run with repo:aaronjmars/aeon --visual to research that repo's last 7 days of commits and PRs, form a falsifiable thesis about its direction, draft a 600β900 word article with cited evidence, generate a hero diagram, and save it to output/articles/. ## Quick Start Ask the AI to write an article using the article skill with an input like "lens:unix philosophy --visual" or "repo:owner/repo" to produce and save a fully cited article.