Humanizer (brandonwise) skill
humanizer
A skill and Node.js CLI that score English text for AI-writing patterns and statistics and suggest fixes on a 0 to 100 scale
- 132 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The `--autofix` flag applies automatic replacements to the text, so run it on a copy of the file
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/brandonwise-humanizerDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The `--autofix` flag applies automatic replacements to the text, so run it on a copy of the file
This is third-party code. Review the repository files before installing.
What it does
Humanizer looks for signs of machine writing using Wikipedia's "Signs of AI writing" list and Copyleaks research. The `SKILL.md` skill installs into OpenClaw and tells the agent how to rewrite a paragraph. The same repository also holds a Node.js `humanizer` CLI, an MCP server and an HTTP API server. The score has two parts. Patterns carry 70% of the weight: detectors for stock phrasing and a vocabulary of over 500 words in three tiers. Uniformity carries 30%: sentence length variation, vocabulary diversity, trigram repetition and paragraph similarity. `humanizer score` returns a number from 0 to 100, where higher means more machine signs. `analyze` shows the matches, `suggest` groups advice by priority, and `humanize --autofix` applies safe replacements. `scan` checks a whole folder and can fail CI above a threshold, while `compare` shows the score difference between two draft versions.
Who it is for. Editors and documentation developers who prepare English texts and want to check drafts and whole `docs/` folders in CI.
Good fit when
- A draft article sounds formulaic and you need a list of spots that give away machine writing
- A documentation folder has to be scanned to find the highest-scoring files
- CI should catch text getting worse against a saved baseline score
- Technical text with code must not be penalized for samples, and the `--ignore-code` mode skips code blocks
- Two draft versions need comparing by number before and after edits
Not a fit when
- The text is in Russian: the dictionaries and patterns are built for English
- You need a verdict from an outside AI-text detector: the score comes from the tool's own heuristics
- Editing legal or academic text: passive voice and formal phrasing belong to the genre there
Example request
Score `draft.md` with `humanizer analyze`, show the ten most visible spots and rewrite the three worst paragraphsLimitations
The dictionaries and patterns are built for English text. The README names 30 patterns and over 500 words while `SKILL.md` says 28 patterns and over 560 words, so exact numbers differ between files. The score rests on heuristics: every analysis carries a confidence rating, and short texts give shaky results. The `--autofix` replacements touch only safe spots and the agent rewrites the rest by hand. The CLI needs Node.js 18. The skill is meant for OpenClaw, and for other agents `SKILL.md` is copied manually.
How to disable. Delete the `~/.config/openclaw/skills/humanizer.md` file and, if you installed the CLI globally, run `npm uninstall -g humanizer`. Delete the cloned repository folder separately.
FAQ
How do I install Humanizer in OpenClaw?
Clone the repository and run `cp humanizer/SKILL.md ~/.config/openclaw/skills/humanizer.md`. The skill needs no package manager.
What does the 0 to 100 number mean?
It is a composite score: the higher it is, the more signs of machine writing the text has. It combines patterns and text uniformity statistics, and no outside service gives a verdict.
Can I run the check in CI?
Yes, `humanizer scan docs --ext md,txt --fail-above 45` fails above the threshold, and the `--baseline` flag together with `--fail-on-regression` catches only regressions.
How does this skill differ from the blader one?
The repository builds on the same Wikipedia list and credits blader/humanizer, but adds a standalone CLI, uniformity statistics, folder scanning and MCP and HTTP servers.
How do I avoid penalizing technical text for code samples?
Add the `--ignore-code` flag and code blocks are skipped. The `--ignore-quotes` flag skips quotes in the same way so pasted examples do not skew the score.
Discussion
No questions or reviews yet.
Related
SkillsA skill that finds AI writing tells, rewrites the text in a chosen voice, and optionally scores it from 0 to 100
Two skills for English prose: humanize rewrites a model draft in a natural voice, and ai-check flags machine-writing tells
A skill that removes AI tone from work writing while keeping every fact, commitment and caveat at the same strength.
An editor for AI-written Russian text: removes bureaucratese, cliches and 38 AI patterns while keeping facts intact
A skill that edits AI-written English in two passes: it removes 26 AI tells and simplifies sentences after ASD-STE100
The Chinese version of Humanizer: finds and removes signs of AI writing in Chinese text
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Humanizer (brandonwise) (this card) | no | 132 | |
| Humanizer | no | 68,355 | a skill built on the same Wikipedia list, while this adds a CLI with a numeric score, folder scanning and MCP and HTTP servers |
| humanizer | no | 311 | another skill on the same topic, while this has uniformity statistics and a version comparison mode |
README in short
The README explains that the project works as an OpenClaw skill and as a standalone CLI and shows three install routes. It goes on with the scoring diagram, a table of statistical metrics, a command reference and newer features: repository scanning, modes that skip code and quotes, version comparison and confidence calibration.
README badge
Original SKILL.md
--- name: humanizer version: 2.2.0 description: > Humanize AI-generated text by detecting and removing patterns typical of LLM output. Rewrites text to sound natural, specific, and human. Uses 28 pattern detectors, 560+ AI vocabulary terms across 3 tiers, and statistical analysis (burstiness, type-token ratio, readability) for comprehensive detection. Use when asked to humanize text, de-AI writing, make content sound more natural/human, review writing for AI patterns, score text for AI detection, or improve AI-generated drafts. Covers content, language, style, communication, and filler categories. license: MIT --- # Humanizer: remove AI writing patterns (v2.2) You are a writing editor that identifies and removes signs of AI-generated text. Your goal: make writing sound like a specific human wrote it, not like it was extruded from a language model. Based on [Wikipedia:Signs of AI writing](https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing), Copyleaks stylometric research, and real-world pattern analysis. ## Your task When given text to humanize: 1. Scan for the 28 patterns below 2. Check statistical indicators (burstiness, vocabulary diversity, sentence uniformity) 3. Rewrite problematic sections with natural alternatives 4. Preserve the core meaning 5. Match the intended tone (formal, casual, technical) 6. Add actual personality — sterile text is just as obvious as slop ## Quick reference: the 28 patterns | # | Pattern | Category | What to watch for | |---|---------|----------|-------------------| | 1 | Significance inflation | Content | "marking a pivotal moment in the evolution of..." | | 2 | Notability name-dropping | Content | Listing media outlets without specific claims | | 3 | Superficial -ing analyses | Content | "...showcasing... reflecting... highlighting..." | | 4 | Promotional language | Content | "nestled", "breathtaking", "stunning", "renowned" | | 5 | Vague attributions | Content | "Experts believe", "Studies show", "Industry reports" | | 6 | Formulaic challenges | Content | "Despite challenges... continues to thrive" | | 7 | AI vocabulary (500+ words) | Language | "delve", "tapestry", "landscape", "showcase", "seamless" | | 8 | Copula avoidance | Language | "serves as", "boasts", "features" instead of "is", "has" | | 9 | Negative parallelisms | Language | "It's not just X, it's Y" |