humanizer-stack

soundshuman is an open-source writing-analysis and repository-auditing toolkit for identifying patterns associated with AI-generated prose. It combines a humanizer skill, a machine-readable rule pack, and a zero-dependency CLI: the skill uses a 41-pattern catalog, statistical signals, voice calibration from writing samples, and a draft–audit–final workflow; the rule pack stores vocabulary tiers, phrase fixes, and weighted regular-expression detectors in JSON; and the CLI scores text from 0 to 100, explains findings with line numbers and triggered rules, applies selected mechanical fixes, scans repositories, and can gate CI. The project states that its goal is inspectable editing rather than defeating AI detectors, with rewrites delivered as reviewable Git diffs and a no-invented-facts rule. The CLI runs on Node 18 or later without dependencies and supports pasted text, files, repository audits, and directory scans. The skill is distributed as plain Markdown for agent harnesses and as a Claude Code plugin; the repository also includes reference materials, a pre-commit hook, CI integration, and installation instructions. It is MIT licensed, with upstream copyrights preserved in the license.

A local command-line reference and checker for common signs of machine-written prose. It provides before-and-after examples, draft scanning, searchable patterns, and an editing guide.

Markdown-based agent skill that rewrites AI-sounding text to read human-written without changing what it says — works with any agent that supports skills. Rewrites against the 35 patterns from Wikipedia's 'Signs of AI writing' (WikiProject AI Cleanup): a first pass free to restructure, then a check of the draft against those patterns and the original claims before rewriting what still sounds artificial. Its rules bar invented facts — names, numbers, dates and quotes must come from the source or the writer — and preserve a writer's personal style (or follow a provided sample). The skill shows its work: the first rewrite and a critique of what still sounds artificial precede the final version.

Open-source Python toolkit that rewrites AI-generated text to read human-written. Ships four documented methodologies — translation chaining, multi-turn LLM rewriting, detection-guided feedback loops, and mixed-engine translation — plus a Standard Pipeline that chains two LLM rewrite passes (DeepSeek/OpenRouter, second pass carries the first as history) with two cross-engine translation hops (EN → ZH → JA → FI → EN across different engines), so no single engine's structural fingerprint survives. Claims 100% key-information retention on 50 verified text pairs; the repo warns detector scores are probabilistic and against misrepresenting authorship.

gstack is an open-source skill and tooling framework for Claude Code, developed by Garry Tan. It turns Claude Code into a virtual engineering team through Markdown-based slash commands and specialist roles covering product review, engineering management, design review, code review, browser-based QA, security auditing, release management, documentation, and other software-delivery tasks. Its tools can drive a real browser and support unit-test, end-to-end-test, and black-box verification loops. The repository describes 23 specialist tools and eight power tools, and distributes the project under the MIT license. It requires Claude Code, Git, and Bun; Node.js is additionally required on Windows.