Treat an AI-text detector as a fitness function for rewriting, with a deterministic claims-diff that flags fabricated facts against the source. Zero runtime deps, shell-first, for Claude Code and Codex.
Universal, zero-dependency LLM skill that strips multi-vendor AI provenance marks (invisible Unicode, exotic spaces, citation/UTM tags, metadata) and transforms robotic AI drafts into authentic, high-variance human prose across any language or genre using a 6-pass stylometric protocol that neutralizes 38 universal LLM signatures.
To associate your repository with the ai-text-analysis topic, visit your repo's landing page and select "manage topics."