No AI Slop
petergyang/no-ai-slop
Removes 20+ patterns of AI-sounding writing while keeping your own voice.
No AI Slop is an MIT-licensed writing skill by Peter Yang that edits a draft to remove more than 20 AI patterns, such as binary contrasts, throat-clearing openers and fake-profound endings, without flattening the writer's voice. It has two jobs: edit a draft and list what changed, or detect patterns and quote each one without rewriting. It also ships as a ChatGPT and Codex plugin.
Facts checked on against the sources listed at the end of this page.
- GitHub stars
- 11,753
- Skills in the pack
- 1
- Agents supported
- 3Also available as a plugin in ChatGPT
- Instruction size
- 1.7k wordsacross its SKILL.md file
How to install No AI Slop
| Agent | Command | Copy |
|---|---|---|
| Any agent | The README also says you can paste an install request into ChatGPT, Claude Code or Codex and let the agent do it. |
Commands as given in the README on 3 October 2026.
What is inside No AI Slop
| Skill or command | What it makes the agent do |
|---|---|
| no-ai-slop | One skill with an editing mode and a detection mode, plus an eval.md file of checks it runs on its own work. |
The same skill covers editing and detecting. Run /no-ai-slop with your text to edit, or ask whether a draft is slop to get a list of quoted patterns.
What No AI Slop runs on your machine
| Hooks | None found in the repo. |
|---|---|
| Bundled scripts | None that run in your agent. A Python build script in the repo packages the ChatGPT and Codex plugin. |
| Network calls | None named in the README. |
| API keys needed | None named in the README. |
These are facts read from the repo, not a security rating. Read the files before you install any skill.
How much context No AI Slop uses
| The skill | 1,715 words in its SKILL.md file |
|---|---|
| Loaded at start | Only each skill's name and description. The full text loads when a task needs that skill. |
Word counts are measured from the repo every day. Reference files a skill loads on demand are not counted.
How No AI Slop works
Edit mode makes the minimum effective edit, returns the draft and adds a What changed section. Detect mode quotes each pattern it finds and gives the fix in a few words. It does not score the draft or guess whether an AI wrote it, since the skill says detectors only guess.
The patterns include binary contrasts, throat-clearing openers, faux-insight setups, colon reveals, dramatic fragments, importance puffery, weasel attribution and synonym cycling. The skill also checks fundamentals: lead with the point when that helps, use active voice, and prefer concrete details over abstractions. A satire mode writes deliberately sloppy AI posts for fun.
A good fit for
- Writers who edit with AI and want to keep their own cadence and humor
- Newsletter authors and bloggers who want a quick check of a draft before publishing
- Editors who want a list of quoted patterns without a rewrite
- Teams that use ChatGPT, Claude Code or Codex, since all three are supported
Skip it if
- Proving that text was or was not written by an AI, since the skill says it does not guess
- Code or technical documentation checks, because it targets prose patterns
- Anyone who wants a full rewrite in a new style, as it makes minimal edits
Is No AI Slop still maintained?
| Last commit | , 1 month ago |
|---|---|
| Latest release | v1.0.6, 1 August 2026 |
| Stars gained in 7 days | Shown after 7 days of tracking |
| Stars gained in 30 days | Shown after 30 days of tracking |
| Open issues and pull requests | 22 |
| Forks | 804 |
Tools No AI Slop works with
Questions about No AI Slop
How do I use No AI Slop?
Run /no-ai-slop followed by your draft to edit it, or /no-ai-slop is this slop? followed by the draft to get a list of patterns without a rewrite.
Is it free?
Yes. It is MIT licensed.
Will it change my voice?
It is built to keep it. The skill tells the agent to notice your vocabulary, cadence and humor, and to leave strong human sentences alone.
Does it tell me if AI wrote my text?
No. Detect mode only quotes the slop patterns it finds. It does not guess whether an AI wrote the text.
Does it work in ChatGPT?
Yes. The README says it is also available as a plugin in ChatGPT, and the repo includes Codex plugin metadata.