Humanizer
Removes signs of AI-generated writing from text using Wikipedia editor patterns.
An agent skill that rewrites predictable AI prose by targeting 26 specific structural patterns.
Humanizer strips common AI writing tropes from text while preserving factual claims and source meaning. It operates as an agent skill within Claude Code, Codex, and other supported developer environments.
What it is
The tool reads full text passages, identifies 26 specific artificial writing patterns across four categories (such as forced triads, staged run-ups, and inflated significance), and rewrites the prose. It shows its work by outputting an initial rewrite, a critique, and a final version.
Who it's for
It suits content writers, editors, and localization professionals who want to eliminate predictable AI phrasing from drafts. It does not suit those looking to evade AI detection tools, as detectors will still flag most of its output.
Key features
Targets 26 AI Patterns
Removes common structural tells like 'Not X but Y', forced triads, and one-line closers.
Voice Matching Support
Matches your specific writing rhythm and quirks when you supply a 2-3 paragraph sample.
Transparent Rewrite Process
Displays the first rewrite, a short critique, and the final version when processing text.
Preserves Structural Integrity
Leaves code, data, frontmatter, and link targets untouched when pointing at a file.
Up and running in 3 steps
Our honest verdict
What's great
- Eliminates repetitive AI phrasing like 'delve', 'testament', and 'landscape'
- Keeps every factual claim and opinion intact from the original text
- Integrates directly into agent workflows like Claude Code and Codex
- Allows voice matching using your own writing samples
Where it falls short
- Does not bypass AI detectors and detectors will still flag the output
- Requires an underlying AI coding agent or compatible CLI environment to run
- Editing quality depends heavily on the capabilities of the host agent
Pricing & license
Free under its open source licence, with the only cost being the execution environment or agent you run it through.
Open remote AI-training roles
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does Humanizer bypass AI detectors?
No. Humanizer edits strictly for human readers, and AI detectors will still flag most of its output.
Which AI agents are supported?
It runs in Claude Code, Codex, pi, Claude.ai, Claude Desktop, and via Skills CLI in Gemini CLI, GitHub Copilot, and Windsurf.
