Dev open-sources Litmus, a personal AI writing detector trained on your own samples
chaseleantj · x · 2026-09-26
Developer chaseleantj released Litmus, an open-source tool that scores whether text sounds like you or like AI.
- It learns your personal writing style from pairs of texts: an AI draft and your own version (minimum 2 pairs).
- Each input gets a score — positive means it reads like you, negative means it reads like a machine — with per-sentence shading to show which lines drift toward AI style.
- It also supports side-by-side comparison of two texts and a 2D map of your entire text library.
Testing it against his own samples, the author found Opus 5.5 sounds about as human as Opus 4.6, while Sol still sounds the most human. Code is on GitHub, powered via OpenRouter, and runs locally with a few commands.
Related event: Litmus: Open-Source AI Detector Trained on Your Own Writing Style(2 posts)→
More from coding & agent
- Matt Pocock: your CODING_STANDARDS.md should be empty for only 5 minutes — mattpocockuk · 2026-09-26
- Custom benchmark: 35B Qwen3.6 scores 95% vs 53% for 120B GPT-OSS on coding agent — pauliusztin · 2026-09-26
- Xiaomi open-sources MiMo RL environments repo, individual tasks sell for $100-1000 each — HarveenChadha · 2026-09-26
- Installing DeepSeek Harness on a Muse cloud VM to unlock Chinese web search — op7418 · 2026-09-26
- Coded animation lands in a single AI one-shot, no assets or examples needed — eschadiol · 2026-09-26
- AI Show & Tell at GitHub HQ: DSPy maintainer, Scalekit CEO on agents — dbreunig · 2026-09-26