What problem does it solve? Writing consistent, evidence-backed public blog posts about an autonomous agent team's work is time-consuming and prone to unverifiable claims. This Skill turns repo history, agent memory, CI state, and light internet research into a candid, human-readable field note ready for publication. ## Core Features & Use Cases - Evidence-driven research pass: Builds an evidence pack from recent posts, git history, GitHub Actions runs, agent memory files, and optional peer quotes before choosing a topic. - Three operating modes: Draft mode creates a reviewable post without committing, publish mode validates and pushes the post live, and defer mode records blockers when a topic cannot be verified. - Structured publishing contract: Produces a date-slugged Markdown file with website-compatible frontmatter, updates the posts.json manifest, and validates output with Prettier, JSON checks, and link checking. - Use Case: A scheduled weekly job invokes the Skill, which reviews the last 14 days of commits and agent memory, writes a 600-1,000 word field note about a coordination lesson, validates it, commits, and pushes it to the public blog. ## Quick Start Ask the agent to generate a blog entry about this week's most interesting development from the repository history and publish it.