Rewrites AI-sounding text so it reads like a person wrote it, without changing what it says or making anything up.
by blader· WritingUnclaimed · Yours? Claim it
4 images
About
Humanizer is a style editor for prose, packaged as an Agent Skill. It finds the habits that make text read as machine-written and rewrites them, keeping every claim and adding nothing new. It's built on Wikipedia's "Signs of AI writing", the guide Wikipedia's editors use, and checks 26 patterns, strongest first. What you can do with it - Paste text after /humanizer, or just ask your agent to humanize it - Rewrite the prose in a file, such as docs/launch-post.md, leaving code, commands, paths, metadata and link targets untouched - Match your own voice by giving it two or three paragraphs of your writing - Use it inside other work: pull request descriptions, commit messages and documents get back only the final text What it looks for - Staging: "not X but Y" contrasts, one-line closers, sayings that sound deep, run-ups like "Here's the thing" - Rhythm by rule: forced triads, repeated openings, dashes everywhere, stacked qualifiers - Inflation: stock AI words, inflated significance, sales language, borrowed authority - Formatting by rule: decorative bold, title-case headings and emojis - Chat leftovers: "Great question!", knowledge-limit disclaimers, and writing about the document instead of its subject How it works For pasted text it shows a first draft, a short list of what still sounds artificial, and the final rewrite. It never invents a name, number, date, quote or citation; if a sentence needs a detail it doesn't have, it asks. Personal writing keeps the writer's opinions and quirks, while technical and reference text stays neutral and plain. How to install it In Claude Code, add the blader/humanizer marketplace and install humanizer@humanizer (Claude Code 2.1.142 or newer). Codex and other agents install it with npx skills add blader/humanizer. In Claude.ai and Claude Desktop, download the repository as a ZIP and upload it as a skill in Settings. Good to know - It edits for human readers. Its maker says getting past AI detectors is not a goal, and detectors still flag most of its output. - A community project by Siqi Chen (blader on GitHub), MIT licensed.
Once it’s added
“Humanize this text: [paste your text]”
“Humanize the prose in docs/launch-post.md.”
“Here's a sample of my writing. Now humanize this post in my voice.”
“Check this draft for signs of AI writing.”
“Rewrite this pull request description so it reads plainly.”
Before you add it
Reads only
The text you paste, or the file you name · A sample of your writing, when you give one
Makes changes · your agent can ask first
The prose in a file you name, leaving code, paths and links as they are
From people who use it
Step by step
Nothing new reaches your agent until a person has looked at it.
Skills people install alongside Humanizer.
No reviews yet. Used Humanizer? Say how it went.
Write a reviewRuns on your machine.
Runs on your machine.
Runs on your machine.
Runs on your machine.
npx -y degit blader/humanizer ~/.copilot/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.cursor/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.claude/skills/humanizer
/plugin marketplace add blader/humanizer /plugin install humanizer@humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.agents/skills/humanizer
Runs on your machine.
Runs on your machine.
npx -y degit blader/humanizer ~/.gemini/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer .agents/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.openclaw/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.config/opencode/skills/humanizer
Runs on your machine.
npx -y degit blader/humanizer ~/.hermes/skills/humanizer
Scanned and passed · checked 3 h ago
Scanned on every change, and again when our rules improve.
GeminiSupported
GitHub CopilotSupportedTell us and a person on the trust team will look at it. Or email [email protected].
Nobody has shared how they use Humanizer yet.
Security scan
Passed
Safe to install in every agent it supports. Runs in your agent, as instructions.
Skills people install alongside Humanizer.
Two documentation skills: HADS, a tagging convention people and AI both read well, and grounded vaults that trace every claim to its source.
Tw93's writing skill: rewrites and polishes Chinese or English prose so it reads naturally, cutting stiff, AI-sounding phrasing.
