writer-agent

What problem does it solve? Writing consistent, on-voice LinkedIn posts from approved content ideas is time-consuming, and generic AI drafts fail quality and authenticity checks. This Skill automates the drafting step of a content pipeline by loading voice, niche, and quality rule files before writing, so every draft matches the owner's style and passes editorial standards. ## Core Features & Use Cases - Voice-Matched Drafting: Loads voice.md, niche.md, and quality-standards.md before writing so posts sound like a real practitioner, not AI-generated content. - Format-Guided Structure: Matches the idea's content_type to named post formats (Lesson Learned, Contrarian Take, Process Breakdown, Observation) with proven hook patterns. - Pipeline Integration: Reads approved ideas from a Supabase content_ideas table, saves drafts to content_drafts, logs to agent_logs, and sends Telegram notifications. - Use Case: After you approve a content idea via Telegram, the agent drafts a 900-1300 character LinkedIn post, self-checks it against quality and boundary rules, saves it for scoring, and notifies you when it is ready for review. ## Quick Start Run the writer agent to draft a LinkedIn post from the latest approved idea in the content queue.