Humanize (AI-Tell Remover)
Rewrites AI-sounding text into human-reading prose using countable gates for em dashes, banned words, and sentence-length spread.
What it does
Rewrites flat or AI-generated text so it matches the statistical and stylistic fingerprint of human writing. Instead of vague "make it warmer" advice, it works from the nine signals detectors actually measure: perplexity, burstiness, hedge density, lexical repetition, structural markers, specificity, POS density, punctuation fingerprint, and RLHF assistant voice.
- Seven hard rules: at most one em dash per 300 words (zero under 300), no semicolons, straight quotes only, a banned-vocabulary list (delve, leverage, robust, comprehensive...), no "not just X, it's Y" negation pivots, no preamble or changelog, and enforced sentence-length spread.
- Nine levers: context-specific word choice, burstiness, hedge surgery, structural flattening, specificity anchors, voice and register, non-AI transitions, punctuation normalization, and stripping the helpful-assistant register.
- Pre-output gate: re-read the draft and write literal counts plus quoted hits, then list every sentence's word count to prove the rhythm conditions pass.
- Signal I audit: a per-paragraph checklist for patterns that feel like good writing but read as AI, such as mini-aphorism closers, tricolons, chiasmus, anaphora, and "turns out" pivots.
Who it's for
- Content writers, newsletter authors, and LinkedIn posters who draft with AI and need a human voice
- Professionals cleaning robotic tone out of reports, memos, and proposals
- Writers who want their own voice preserved by supplying samples (protocol step 0 distills style hypotheses)
- Editors auditing drafts for machine-writing habits
Examples
- Polish a draft: "humanize this newsletter section" and get back only the rewrite, with em dashes cut, hedges deleted, and a mix of 5-word fragments and 25-plus-word sentences.
- Match your voice: paste three past posts and ask for a rewrite in that register; the skill extracts 5-10 style hypotheses about sentence variance, openers, and verbal tics first.
- Diagnose: "why does this read as AI?" returns an inventory of hedge count, bullet overuse, sentence uniformity, and RLHF voice markers.
- Specificity warning: if the input has zero numbers, names, or dates, it still rewrites and then appends a note that real specifics are needed to pass the specificity signal.
· · · Install guide · · ·
Try it now, no install
Paste this into Claude to use the skill without installing anything.
Read the instructions in this file and follow them to help me: https://raw.githubusercontent.com/harshaneel/humanize/HEAD/humanize/SKILL.md What I want: (describe your task here)
If Claude can't open the link, open it yourself and paste the contents instead.
↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.
Install in the Claude app (no terminal)
- Download the ZIP with the button below.
- In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
- Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
Install in Claude Code
Let Claude do it — paste this into Claude Code
Install the skill I found on Claude Skill Mart. Copy the humanize folder from the GitHub repo harshaneel/humanize into my ~/.claude/skills/humanize/. When it's done, tell me in one line what this skill can do.
Install with a command instead
git clone https://github.com/harshaneel/humanize.git /tmp/humanize && mkdir -p ~/.claude/skills && cp -r /tmp/humanize/humanize ~/.claude/skills/humanize⚠ This is a third-party skill. Check the source repository before installing.
- Open a terminal (Terminal on macOS, WSL or Git Bash on Windows).
- Make sure the skills directory exists:
mkdir -p ~/.claude/skills - Clone the repo to a temp folder:
git clone https://github.com/harshaneel/humanize.git /tmp/humanize - Copy the skill folder into place:
cp -r /tmp/humanize/humanize ~/.claude/skills/humanize - Verify the files:
ls ~/.claude/skills/humanizeshould show SKILL.md and a references folder. - Restart Claude Code, then trigger it with a request like "humanize this paragraph" or "make this sound less AI".
- If the repo ships a companion skill such as
ai-check, copy it the same way to get the grading pass too.