titus-humanizer v1.1.0

Plain rewrites that keep every fact

An editing skill that rewrites text so it reads like a careful, plain-spoken person wrote it. It strips hidden characters first and scores the AI-writing patterns before and after.

Check your text Watch the tour Install

Draft

In today's fast-paced digital landscape, our new onboarding checklist isn't just a tracker. It's a gateway to a seamless experience, fostering alignment across all 12 setup steps. Moreover, it serves as a pivotal tool that empowers new hires and their managers to navigate the first two weeks with confidence. Ultimately, it is a testament to our commitment to people.

--AI-pattern score

AI-pattern score

The highlights and scores are the titus-humanizer v1.1.0 scoring script's own output for these two texts. The rewrite keeps the 12 steps, the two weeks and who uses the list. The score also measures sentence rhythm and word variety, so plain text rarely scores 0.

Four ways to use it

You don't need Claude or ChatGPT. The Checker needs no account at all.

Check any text

Get the 0 to 1 score, every pattern marked in place and the hidden characters removed. Quoted text stays out of the score. It runs in your browser and costs nothing.

Open the Checker

Rewrite with your key

Use a Gemini, Claude or OpenAI key. Google's free tier costs nothing, but Google may use free-tier text to improve its products.

Rewrite a draft

Use any chatbot

Copy the rulebook into Gemini, Copilot, Perplexity or whatever you already use.

Get the rules

Install it

Give Claude, ChatGPT, Claude Code or Codex the repository link and it installs the skill for you. It then loads whenever you ask for a rewrite.

See the install steps

The tour

Every feature in under three minutes

2 minutes 49 seconds, with captions

Read the transcript

Release

Meet titus-humanizer, a new editing skill for Claude and ChatGPT, now published on GitHub.

Rewrite

Give it a draft that sounds like AI, and it marks each pattern that gives the draft away. The stock opener. The em dash. The framing that says it isn't just one thing, it's another. And the clause tacked on at the end. Then it rewrites the sentence in plain words, and keeps every fact.

Score

Each rewrite comes with a score from zero to one, before and after. This sample email from the test suite drops from 0.80 to 0.01, with no patterns left. The score counts writing patterns. It isn't an AI detector.

How it works

Here's how it works. First, it removes hidden characters, such as zero-width spaces. It applies seven hard rules. It fixes wording that's awkward, flowery, vague, unnecessary or illogical. It cuts story-level patterns. And it scores the original and the rewrite.

Stories

It handles fiction, too. Using the StoryScope study of AI-written stories, it cuts weather that mirrors the mood, fear told through body sensations, and morals the narrator spells out. The rewrite keeps her decision and names the feeling plainly.

Formats

Use it on emails, reports, slide decks, résumés, papers and more. Résumé date ranges and bullets come through intact.

Easiest install

The simplest way to install is to copy this repository's link into Claude, Claude Code, ChatGPT, or Codex, and ask it to help you install the skill. It will handle the installation for you, or walk you through any step it can't do itself. To install it yourself, follow the steps that come next.

Install in the app

To install it in the Claude app, first download the zip file from the Releases page on GitHub. In Settings, open Capabilities, and turn on Create and edit files. Then open Customize, then Skills. Click plus, choose Create skill, then Upload a skill, and select the zip. Turn the skill on, and you're ready. In ChatGPT, go to Skills, select Create, and upload the same zip without unzipping it. If your plan doesn't support skills, you can build a Custom GPT from the instructions in the repo.

Install from the command line

In Claude Code, two commands add the repo as a plugin marketplace and install the plugin. For Codex, clone the repo, and copy the skill folder into the agents skills folder in your home directory.

Get it

titus-humanizer works in Claude, Claude Code, ChatGPT and Codex. It's free under the MIT license, and the link is on screen.

How it works

It cleans the text, rewrites what it finds and then measures the result.

Clean

It removes characters you can't see, such as zero-width spaces and look-alike letters, and leaves emoji, other alphabets and code alone.

Rewrite

It applies seven hard rules, fixes five kinds of weak wording and removes the patterns AI stories share. Every fact, number, name and quotation stays as it was.

Measure

It scores the original and the rewrite from 0 to 1, and takes another pass if the rewrite still scores high.

The score counts the AI-writing patterns titus-humanizer targets. It doesn't measure quality. It isn't a detector, and a low score doesn't prove a person wrote the text. The cleanup removes invisible characters only. AI watermarks and file metadata stay in place.

It works on almost anything you write

Emails, reports, slide decks, résumés, papers, cover letters, personal statements, blog posts and fiction.

Seven hard rules

No em dashes or semicolons, no "it's not X, it's Y", no passive voice that hides who acted, no slogans and no chatbot phrases.

Five kinds of weak wording

Awkward, flowery, vague, unnecessary and illogical. Each comes with a test, such as "would you say this out loud to a colleague?"

Story patterns

AI stories state their moral, stack body sensations for fear and make the weather match the mood. The skill cuts these and keeps your real details.

Before and after

Each rewrite is shorter and plainer, with the same facts

None of the rewrites add facts. When a draft is vague, the skill writes the plainest honest sentence.

Announcement

Before

In today's fast-paced digital landscape, our new onboarding checklist isn't just a tracker — it's a gateway to a more seamless experience, fostering alignment and driving meaningful impact.

After

The new onboarding checklist shows new hires which setup steps they have finished.

The report the scoring script printed

AI-pattern score: not scored (28 words, needs at least 30) before, not scored (13 words, needs at least 30) after
Patterns before: AI vocabulary x3, em dash x1, reject-then-assert x1, tacked-on -ing clause x1, stock opener x1
Patterns left after: none
Hidden characters removed: none

Email

Before

I hope this message finds you well! I wanted to reach out to touch base regarding the proposal, which was reviewed by our team last week. It's not just a budget request — it's a roadmap for sustainable growth. Please let me know if you have any questions!

After

Our team reviewed the proposal last week. It sets out the budget we're requesting and what it would cover. I'm happy to walk you through the numbers.

The report the scoring script printed

AI-pattern score: 0.56 (medium) before, not scored (27 words, needs at least 30) after
Patterns before: em dash x1, reject-then-assert x1, passive voice (heuristic) x1, chatbot residue x1
Patterns left after: none
Hidden characters removed: none
Note: the text is short, so the script dampened the score and it is less reliable.

Story

Before

Elena stood at the window as the rain fell, mirroring the storm inside her. Her chest tightened, ice crawled down her spine, and the scent of wet earth and old memories filled her lungs. She realized, at last, that letting go was not the same as forgetting. It was a way of carrying him with her.

After

Elena stood at the window and watched the rain. She was afraid. She decided she could let him go without forgetting him.

The report the scoring script printed

AI-pattern score: 0.60 (high) before, not scored (22 words, needs at least 30) after
Patterns before: emotion as body sensation x2, setting as mirror x1, decorative smell x1
Patterns left after: none
Hidden characters removed: none
Note: the text is short, so the script dampened the score and it is less reliable.

Source code on GitHub. MIT licensed.