lynote-ai

lynote-ai / ai-detector-skill

Public

A free detector capable of identifying content generated by all advanced AI models.

lynote.aiai-detector ai-agents ai-detector awesome-list detector skiils
19
1
85% credibility
Found May 28, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is a free, locally-run tool that analyzes any written text to estimate how much it looks like it was written by AI. It gives you a score between 0-100, explains which patterns it found most suspicious, and constantly reminds you not to treat its output as proof. The tool is designed to help with honest, human-supervised review: teachers spotting suspicious essays, editors flagging formulaic guest posts, or content teams comparing drafts. It refuses to classify very short samples and warns against using it for high-stakes disciplinary decisions. Think of it as a thoughtful assistant that gives you a starting point for conversation, not a judge that declares guilt.

How It Works

1
🔍 Discover the tool

You learn about a free tool that can help you spot AI-written text, designed to stay honest about what it can and cannot tell you.

2
📦 Install it on your computer

You download and set up the program in minutes using a simple command, getting a small assistant that runs entirely on your machine.

3
✍️ Run the analyzer on any text

You paste in a student essay, a product review, or any written sample, and the tool examines it for patterns that AI writing often shows.

4
📊 Get your full report

You receive a detailed breakdown: a risk score, a verdict like 'suspicious' or 'uncertain', the strongest clues the tool found, and clear warnings about when NOT to over-rely on this information.

5
Choose your path based on results
👩‍🏫
Teacher checking homework

Use the findings as a starting point for a conversation, not a punishment. Compare suspicious passages against known writing samples.

📝
Editor reviewing submissions

Identify formulaic or templated prose that feels off, then decide whether to ask for revisions or do deeper editing.

👥
Content team comparing drafts

See how different your team-written version is from an AI-assisted version and decide what level of editing each needs.

6
🧠 Apply your own judgment

The tool gives you a careful nudge in one direction, but you remain in control—you combine the findings with your own reading and context.

✅ Make a thoughtful decision

You have a cautious, explainable signal that helps you decide what needs human review, without overclaiming or treating the tool as proof of anything.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 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-detector-skill?

This is a Python CLI and library that analyzes text for AI-generated writing signals. You feed it a file or pipe text through stdin, and it returns a risk score from 0-100, a verdict like "high_ai_likelihood" or "insufficient_text," and explains which signals drove the decision. The tool runs entirely offline with no network calls, making it suitable for local workflows and CI pipelines. It ships as a skill package ready to drop into agent environments like Claude Code or Codex.

Why is it gaining traction?

The hook is honesty. Most AI detectors overclaim. This one refuses to. It includes a short-text guardrail that bails out on samples under 80 words, and every result comes with caveats reminding you not to use it as proof of misconduct. The evaluation scripts benchmark against the public HC3 dataset and generate reproducible reports, which builds trust. For developers building agent workflows, the skill-ready packaging means you can drop it into an agent's skill directory and get explainable signals instead of a black-box score.

Who should use this?

Teachers doing initial triage of student submissions before manual review. Editors scanning guest posts for formulaic AI prose. Trust and safety teams prioritizing long-form content for human review rather than auto-removal. Content teams comparing drafts to spot where language gets too generic. Not for disciplinary decisions, high-stakes authorship disputes, or samples under 80 words.

Verdict

The credibility score of 0.85% and 19 stars reflect a very early-stage project with limited real-world validation. The conservative design philosophy is refreshing, and the agent integration story is genuinely useful. But this is alpha software without production hardening. Good for experimentation and embedding in agent workflows, not for enforcement decisions.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.