FeedHive AI Slop Detector is a free, MIT-licensed 120-pattern 'slop' catalog plus a Claude Code and Codex skill that lints AI writing before it ships.
Yes, FeedHive AI Slop Detector is open source under the MIT license.
Free, MIT — git clone or consume patterns.json directly; no signup, payment, or API key for the open artifact. (free).
Yes, FeedHive AI Slop Detector can be self-hosted (the source is available under the MIT license).
A free, MIT-licensed library of 120 AI-writing "slop" patterns organized into 9 categories, shipped two ways: as a structured patterns.json you can wire into any tool, and as an agent skill that drops into Claude Code or Codex so an AI model checks its own draft against the catalog before it publishes. Where most "humanize my AI text" products are hosted black boxes, this one is the raw, inspectable rulebook — clone it, read every pattern, change it, and ship it inside your own stack.
AI now writes a large share of GTM content, and the giveaway tells — the "it's not just X, it's Y" cadence, the rule-of-three everywhere, the em-dash pile-ups — are exactly what makes that content read as machine-made and lose trust. FeedHive, a social-media scheduling company, open-sourced the internal checklist it uses to catch those tells, as both a data file and a drop-in agent skill, under MIT. The honest read: it is a pattern catalog, not a quality engine — it strips obvious tells but will not turn weak writing into good writing, and the "most comprehensive list on the internet" line is the vendor's own framing, not a verified fact. It is also a fresh repo with modest traction. But as a free, local, fully inspectable lint pass for an AI content pipeline, it is a cheap, honest piece of leverage.