What problem does it solve? Published SEO articles often fall short of top-tier editorial benchmarks, and manually auditing text, layout, and CTAs against reference sites like NYT Wirecutter is slow and inconsistent. This Skill automates a post-publish editorial loop that audits, reports, and rewrites articles without human intervention. ## Core Features & Use Cases - Autonomous post-publish audit: Fires after every successful publish via ctx.waitUntil, reading article HTML from KV and running a Kimi-based text audit against a per-category reference URL. - Visual screenshot analysis: Drives Composio's BROWSER_TOOL to load, scroll, and screenshot the live page, then audits layout, density, and CTA issues from the screenshots. - A/B-safe rewrite and republish: Writes an improved HTML variant to a -b KV key (preserving affiliate links and product-pick blocks) so a split-tester can compare variants without touching the live article. - Use Case: After the SEO pipeline publishes a new cat litter box roundup, the agent automatically compares it to Wirecutter's benchmark, logs an EditorialReport to KV, and produces a rewritten variant for A/B testing. ## Quick Start Run an editorial audit on the published article at a given KV key by posting to the /api/admin/editorial-review endpoint with the kvKey and an optional reference URL.
