kdgbalmer

kdgbalmer / ai-tells

Public

A Claude skill that catches AI writing tells in any draft and rewrites them clean. Flags vocabulary (delve, tapestry, leverage), structures ("It's not X, it's Y"), tone ("Great question!"), and punctuation (em dash overuse). MIT licensed.

16
1
100% credibility
Found May 08, 2026 at 16 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

ai-tells is a Claude AI companion that scans drafts for common AI-generated writing patterns and rewrites them to sound more authentically human.

How It Works

1
📖 Discover ai-tells

You hear about ai-tells, a handy tool that spots robotic writing patterns in your drafts and makes them sound naturally human.

2
Pick your setup style
📁
Claude workspace

Create a new space in Claude and add the helper file there.

📂
Skills folder

Put the helper file in your Claude skills spot on your computer.

✏️
Custom instructions

Copy the helper details right into your Claude settings.

3
Helper is ready

No matter which way you choose, your writing assistant is now active and waiting for your text.

4
Share your draft

Paste in your writing and say something like 'humanize this' or 'remove the AI feel' to kick things off.

5
🔍 Review the checkup

You get a clear list of robotic spots with fixes, a score on how AI-like it was, and a smooth new version.

🎉 Natural writing win

Your text now flows like it came straight from you, ready to share without any robot vibes.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 16 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ai-tells?

ai-tells is a Claude skill that scans drafts for AI writing tells—like "delve," "tapestry," "leverage," cheesy structures such as "It's not X, it's Y," or em dash overuse—and rewrites them to sound human while keeping your meaning intact. Drop in text via Claude projects, the skills folder, or system prompts; it outputs an audit of flags, a tell-density score per 100 words, and a clean version. Built as a Markdown-based Claude skill from a GitHub repo, it draws from Wikipedia's AI signs list for easy, updatable detection.

Why is it gaining traction?

Unlike generic AI writing tools, it auto-triggers on phrases like "humanize this draft" or "remove the AI smell," integrating seamlessly as a Claude GitHub skill, app, or connector in projects or ~/.claude/skills/. Developers grab Claude skills examples or downloads from releases for quick setup, and the editable catalogue stays fresh against evolving tells, from goblin-era leaks to new ones. It's a lightweight Claude GitHub plugin or MCP that beats bloated detectors by focusing on actionable rewrites.

Who should use this?

Tech writers and bloggers polishing Claude-assisted posts, marketers fixing AI copy for emails or landing pages, and devs drafting natural READMEs or docs. Content teams spotting tells in collaborative drafts, or anyone using Claude GitHub actions, Copilot integrations, or custom skills to humanize output without losing voice.

Verdict

Grab it if you use Claude daily—MIT licensed, dead-simple install, and PR-friendly for updates—but at 16 stars and 1.0% credibility, it's raw with just a README; treat as a starter kit to fork for production. Solid for quick wins, pair with human review for polish.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.