No AI Slop: A Skill That Removes 20+ AI Writing Tells Without Flattening Your Voice
Removes 20+ patterns of AI slop from any piece of writing. That's the whole thing." | It also enforces the fundamentals that make writing good: Lead with the point when it helps, use active voice, untangle hard-to-follow sentences, and prefer concrete numbers over abstractions.
At a glance
- What is it?
- A Python-packaged skill for ChatGPT, Claude Code, and Codex that detects and removes over 20 patterns of AI-flavored writing. It is a rule-based editor, not a detector, and it preserves your personal voice while fixing the fundamentals.
- Who is it for?
- Adopt No AI Slop if you are a writer or editor who uses AI assistants and wants a consistent, rule-based check for common AI tells like binary contrasts and throat-clearing openers. Do not adopt it if you need a probabilistic AI-detection tool or if you want to preserve every stylistic quirk in your drafts, because the skill will flag and rewrite patterns you may consider your own.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 14 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 14, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What No AI Slop Actually Does
No AI Slop is a skill that installs into AI coding agents and chat assistants. It targets a specific annoyance: AI-generated text that reads cleanly but sounds like every other AI-generated text. The README lists examples like "It's not X. It's Y." and "What nobody tells you is..." and calls them slop. The skill checks for over 20 such patterns, from binary contrasts to fake-profound endings. It also enforces writing fundamentals: leading with the point, using active voice, untangling hard-to-follow sentences, and preferring concrete numbers over abstractions. This is not a plagiarism checker or a statistical AI detector. It is a rule-based editor that scans your text for known patterns and rewrites them. The skill's purpose is to remove the tells without stripping out your vocabulary, cadence, humor, or imperfections, which is a harder problem than simply deleting phrases.
How It Works: Rules, Not Probabilities
The mechanism is visible in the repository structure. A SKILL.md file contains the editing rules and workflow, and an eval.md file contains the checks the skill runs on its work. The skill is invoked with a slash command: /no-ai-slop followed by your writing. It then applies the pattern list, which includes binary contrasts, throat-clearing openers, faux-insight setups, colon reveals, dramatic fragments, superficial analysis, importance puffery, weasel attribution, synonym cycling, and fake-profound endings. Each pattern is described in the README with an example. For instance, synonym cycling is flagged when the same concept is described with different words in consecutive sentences, like "The agent handles your email. The assistant drafts replies." The skill does not guess whether AI wrote the text; it quotes every slop pattern it finds. That distinction matters: it is a deterministic editor, not a probabilistic detector. It also has a mode to detect slop, invoked as /no-ai-slop is this slop? (your writing), which lists the patterns found without making a judgment about authorship.
Installation and Usage: Commands That Work Today
You can install No AI Slop in two ways. The easiest is to paste a single sentence into ChatGPT, Claude Code, Codex, or another coding agent: "Install the /no-ai-slop skill globally from https://github.com/petergyang/no-ai-slop". The README also gives an npx command: npx skills add petergyang/no-ai-slop --skill no-ai-slop --global --yes. That command assumes you have Node.js and npx available. Once installed, you invoke the skill with /no-ai-slop followed by your writing to edit it, or /no-ai-slop is this slop? followed by your writing to detect patterns. The skill lists what it changed, which is useful for reviewing the edits. There is also a plugin.json file for ChatGPT and Codex plugin metadata, meaning it can be loaded as a plugin rather than a slash command. The repository includes a build_plugin.py script that builds and validates the plugin package, so you can rebuild it if needed. No Python installation is required for end users, because the skill runs inside the agent's environment, not as a standalone script.
Where It Falls Short: No AI Detection, No Style Guarantee
The biggest limitation is that No AI Slop is not an AI detector. It cannot tell you whether a text was written by a human or a machine. It only flags known patterns. If an AI writes in a style that avoids all 20+ patterns, the skill will pass it. Conversely, a human writer who uses phrases like "Here's the thing" or "It's not X. It's Y" will get flagged and rewritten. That is a trade-off: the skill enforces a specific editorial style, not a universal standard. Another limitation is that the pattern list is fixed and may not cover new or evolving AI slop. The README says it checks 20+ patterns, but it does not list all of them, so you cannot verify coverage without reading the SKILL.md file. The skill also claims to preserve personal voice, but that is a tall order for a rule-based system. Rewriting a dramatic fragment like "That's it. That's the whole thing." necessarily changes the rhythm of the sentence. If your voice relies on such fragments, you will have to manually revert some edits. The README's own example of the skill's output is not shown, so the quality of rewrites is unverified from the material.
The Alternative: Probabilistic AI Detectors
The obvious alternative is a statistical AI detector like GPTZero or Originality.ai. Those tools use machine learning to estimate the probability that a text was AI-generated. They do not rewrite anything; they just give a score. No AI Slop takes the opposite approach: it rewrites, but it does not judge authorship. The difference in approach is fundamental. A detector answers the question "Did an AI write this?" with a probability. No AI Slop answers the question "Does this text contain patterns I want to remove?" with a list of edits. Which one you need depends on your goal. If you are an editor trying to clean up AI-assisted drafts, No AI Slop is more useful because it acts on the text. If you are a teacher checking for plagiarism or AI use, a detector gives you a score, but it cannot tell you what to change. No AI Slop also differs from style guides like Strunk and White, which are static rules. This skill is dynamic: it runs inside an AI agent and can adapt to the specific text you feed it, but it is still bound to its pattern list.
Maintenance, License, and Upgrade Cost
The repository is actively maintained. The last push was on 2026-08-01, and recent releases include v1.0.6, v1.0.5, and v1.0.4, with dates spanning July 27 to August 1, 2026. That suggests frequent updates, likely to refine the pattern list. The license is MIT, which means you can use, modify, and distribute the skill freely, including in commercial products. There are no restrictions on redistribution as long as the MIT license text is included. The upgrade cost is low: because the skill is installed as a global skill, you can update it by re-running the npx command or asking the agent to reinstall it. The README does not mention a version pinning mechanism, so updates may change behavior without notice. If you rely on a specific set of patterns, you should check the release notes before updating. The repository also includes an eval.md file, which suggests the author runs checks on the skill's output, but the material does not describe what those checks cover or how rigorous they are. You should treat the pattern list as a starting point, not a complete solution.
Editorial conclusion
Adopt No AI Slop if you are a writer or editor who uses AI assistants and wants a consistent, rule-based check for common AI tells like binary contrasts and throat-clearing openers. Do not adopt it if you need a probabilistic AI-detection tool or if you want to preserve every stylistic quirk in your drafts, because the skill will flag and rewrite patterns you may consider your own. Before relying on it, verify that the pattern list matches the kind of writing you produce, and test it on a sample of your own text to see whether the rewrites align with your voice. The skill is MIT-licensed and actively maintained, but it is a set of rules, not a guarantee of originality.
Community notes