AI Humanizer
humanize
Open-source, research-grounded skill suite for humanizing AI-generated text and detecting AI writing, with nine stylometric levers derived from 50+ peer-reviewed sources.
humanize
What is humanize?
humanize is an MIT-licensed, rule-based skill suite that provides an AI text humanizer (`humanize`) and an AI-content detection scorer (`ai-check`), designed to be installed as static skills in LLM agents like Claude Code, ChatGPT, and Codex CLI.
humanize vs Similar AI Tools
| Pricing Model | Free | Free | Free | Free, Free Trial, Paid |
| Free Credits | ||||
| Key Features |
|
|
|
|
| Pros |
|
|
|
|
| Cons |
|
|
|
|
| Best For |
|
|
|
|
How to use humanize?
- 1Clone the repository from GitHub.
- 2Run the install script (`./install.sh all`) to install both skills into your LLM agent.
- 3Use the `/humanize` command followed by text to rewrite it as human-like, or `/ai-check` to get a forensic analysis.
- 4Alternatively, invoke skills by natural language prompts like 'humanize this' or 'does this sound AI?'
humanize Key Features
- Two skills: humanize (rewrites AI text to appear human) and ai-check (forensic AI detection scoring)
- Nine stylometric humanization levers addressing perplexity, burstiness, hedges, structure, specificity, voice, discourse, punctuation, and RLHF artifacts
- Works with any LLM agent (Claude Code, ChatGPT, Codex CLI, Gemini, Cursor, etc.)
- Static rule-based skills, no runtime dependencies or API calls
- Grounded in 50+ peer-reviewed sources through April 2026
- Supports voice matching via writing samples for personalized outputs
- Open-source under MIT license
humanize Use Cases
- Making AI-generated academic essays or articles appear human for basic detector evasion
- Auditing content for AI authorship evidence with detailed scoring and quoted flags
- Ensuring platform trust by verifying human or AI origin of text
- Combining both skills: first detect AI tells, then humanize to address specific flags
humanize Pricing & Free Credits
humanize currently operates on a Free model.
This tool is completely free to use
humanize Pros & Cons
Pros
- Beats perplexity-based detectors (ZeroGPT, QuillBot) in benchmarks
- Research-backed with 50+ peer-reviewed sources
- Works in major LLM agents with simple installation
- Static and private (no data sent to external servers)
- Open-source MIT license allows customization and auditing
Cons
- Does not defeat learned commercial classifiers like GPTZero or Grammarly
- Rule-based, so not adaptive to new detection methods without updates
- Requires agent integration, no standalone UI
- May produce false confidence if used naively against advanced detectors
What is humanize best for?
- Making AI-generated text appear human for bypassing basic detectors
- Checking text for AI generation signals with evidence quotes
- Developers and researchers needing a customizable, open-source humanization tool