Review every draft (x, linkedin, devto) against extracted_insights.md. Kill generic claims, fix accuracy, enforce platform fit. This is the verification step a single LLM call lacks. Writes review_notes.md and corrects the drafts in place.
Write a 1000-1500 word dev.to technical article from extracted_insights.md following Problem -> What I built -> How it works -> What I learned -> What's next. Honest about tradeoffs. Saves devto_draft.md.
Extract the real, source-grounded technical insights from a developer's README, learnings, and hard-parts. Use BEFORE any drafting so writers only work from verified material. Produces extracted_insights.md and never writes platform content itself.
Write a 150-300 word LinkedIn post from extracted_insights.md. Personal narrative, short paragraphs, what broke and what was learned. No corporate tone. Saves linkedin_draft.md.
Write a platform-native X (Twitter) thread of 6-10 tweets from extracted_insights.md. Hook-first, no "I built X" openers, every claim traceable to the source. Saves x_draft.md.