sepia

Why we included this project

Most humanizer tools stop at swapping words and smoothing syntax, but sepia works on the structural patterns that actually give AI prose away. It's a portable Agent Skill that runs inside Claude Code, Codex, Grok Build, and Antigravity, with four operations: write, review, refactor, and recreate. Review only diagnoses a draft, refactor makes minimal edits, and recreate does a full rewrite, so you can choose how much intervention you want. Fiction writers get a three-pass protocol built on the StoryScope research, which found that classifiers detect narrative architecture (explained themes, tidy causal chains, linear time) far more reliably than surface style. Professional writers get venue-matched rule files for release notes, PR replies, postmortems, tickets, and technical articles, each with its own checklist. If you regularly ship AI-assisted text and want it to read like a human wrote it, this is a research-backed alternative to generic humanizer prompts.