no-ai-slop skill: edit AI-sounding drafts and keep your own voice

Peter Yang's skill that makes your agent edit a draft into sharper writing, or list the AI-sounding patterns in it, while keeping your own voice.

Installs: 29.2K; on skills.sh
Stars: 13K; on GitHub
Files: 2; besides SKILL.md
Updated: Sep 2026; last repo commit

no-ai-slop has two jobs. Edit mode, the default, returns your draft with the minimum changes needed and a short "What changed" list. Detect mode names each AI pattern it finds, quotes the line and gives the fix in a few words, without rewriting or guessing whether AI wrote the text.

It loads when you ask for a draft to be clearer, more direct or less AI-sounding, or ask if writing reads as AI. It asks you to paste a draft if you did not. If the audience or goal is unclear it asks one question. It bundles no scripts and SKILL.md names no tools: its checks are written rules, a banned-word list and 18 named patterns.

Good for

Writers and founders who draft with AI and want the result to sound like them
Editors who want an AI-pattern check that quotes each line instead of guessing who wrote it
Newsletter and social authors cleaning up posts before publishing

Install no-ai-slop

Claude Code

Copy this prompt.

Install the no-ai-slop skill in Claude Code. Run this command:

npx skills add https://github.com/petergyang/no-ai-slop --skill no-ai-slop

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Edit drafts into sharper, more human writing while preserving the writer's personal voice, or detect AI-slop patterns without rewriting. Use when the user wants a draft clearer, more direct, more opinionated, or less AI-sounding, or asks whether writing reads as AI.

Sections in SKILL.md

Two jobs
What to ask for
Editing principles
Words to cut
Patterns to cut
Workflow

Key facts

Author: Peter Yang (petergyang/no-ai-slop repo)
Installs on skills.sh: 29,238 as of 2026-11-02
GitHub stars: 13,020 (2026-11-02)
License: MIT (stated at the end of the README)
Last repo commit: 2026-09-02
Files: 3 files including SKILL.md: eval.md (self-check list) and agents/openai.yaml; no scripts
Works in: README says to install it from ChatGPT, Claude Code, Codex or another coding agent, and lists a ChatGPT plugin

What it includes

Returns the edited draft plus a What changed section
Makes the smallest edit that fixes AI patterns, errors, repetition and unclear passages. Strong human sentences are left alone.
Try: “/no-ai-slop Edit this LinkedIn post and keep my voice: (paste draft)”
Read · Edit (default)
Quotes each AI pattern with a short fix, no rewrite
Lists each pattern from the skill that appears, quotes the line and gives the fix. It does not score the draft or guess if AI wrote it, and offers to edit afterwards.
Try: “/no-ai-slop is this slop? (paste draft)”
Read · Detect
Asks for a draft, the audience, or the reader goal if missing
Read · Questions before editing
A banned-word list, plus empty adverbs and filler phrases
Banned outright: words like delve, leverage, robust and tapestry. Empty adverbs and phrases are cut unless they carry emphasis or the writer's own rhythm.
Read · Words to cut
18 named patterns, from binary contrasts to recap endings
Includes binary contrasts ("not X, it's Y"), throat-clearing openers, faux-insight setups, colon reveals, weasel attribution, fake-profound endings, formatting slop and em dashes.
Read · Patterns to cut
Runs a pass/fail checklist on its own edit before replying
After an edit, the agent checks the draft against eval.md and fixes anything that fails before it returns the result.
Read · eval.md self-check
README prompt that writes satirical AI slop on request
The README says a prompt like "Draft an AI slop post about (topic)" produces deliberately bad AI-style writing as satire.
Read · Generate slop for fun

From the official docs

Questions

What does no-ai-slop do?

It edits a draft to remove AI-sounding patterns while keeping your voice, or in detect mode lists the patterns it finds with quoted lines and short fixes. In edit mode you also get a "What changed" section.

How do I install it in Claude Code, the Claude app or Codex?

Use the install box at the top of this page. The README also says you can paste "Install the /no-ai-slop skill globally from https://github.com/petergyang/no-ai-slop" into ChatGPT, Claude Code, Codex or another coding agent.

Does it need an API key?

No key is mentioned in SKILL.md or the README. The skill bundles no scripts; it is a set of editing rules for your agent.

Is it safe?

It bundles no scripts and SKILL.md names no tools to run. You paste your draft into the chat, so it goes wherever your agent sends your messages. Check the edit against your original before you publish, since it rewrites your text.

Will it flatten my writing style?

That is what the skill is built to avoid. SKILL.md tells the agent to keep your vocabulary, cadence, bluntness and humor, make the minimum effective edit, and leave strong human sentences alone.

Can it tell me whether AI wrote my text?

No. SKILL.md says AI detectors guess, so detect mode only names patterns and quotes them as evidence you can check. It does not score the draft or claim AI authorship.

Information

updated Nov 2, 2026

Repository: petergyang/no-ai-slop
Folder: skills/no-ai-slop

petergyang/no-ai-slop skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

no-ai-slop SKILL.md · github.com
no-ai-slop eval.md · github.com
no-ai-slop README · github.com
skills.sh listing · skills.sh