Skip to content
Get packs

Stop Slop

A set of writing rules that helps Claude or other LLMs remove common AI-style phrasing from prose.

GitHub Download source
Stars
18.5k
Forks
1.4k
Licence
MIT
Language
—
Last update
8 months ago
Type
AI model / agent

What it is

Stop Slop is a skill file, a set of instructions, that teaches an AI model such as Claude to spot and remove predictable patterns in writing, like stock phrases, stiff sentence structures, and repetitive rhythms. The README describes it as a core SKILL.md plus reference files listing phrases to remove, structures to avoid, and before-and-after examples. People use it to make AI-assisted writing sound less generic.

  • Adding the folder as a skill in Claude Code so the model applies the rules when drafting
  • Uploading SKILL.md and reference files to a Claude Project's knowledge
  • Copying core rules from SKILL.md into custom instructions
  • Including SKILL.md in a system prompt for API calls, with reference files loaded on demand

Great for

  • People who use Claude or another LLM to draft text and want output that reads less formulaic
  • Writers who want a checklist of banned phrases and structural habits to apply during editing
  • Developers who want to load a writing style guide into a system prompt or Claude Project

Not ideal for

  • Anyone expecting a standalone app or editor; the README describes only instruction files, not a tool with its own interface
  • Users who need guaranteed detection of AI-generated text, since the project is about style rules, not detection
  • Teams needing a style guide for a specific publication, because the rules are general and the README does not say how to adapt them
Step-by-step guide: how to use Stop SlopInstall, first run, examples, tips and FAQs

How to get it

  1. 1

    Clone and build from source

    Clone the repository

    git clone https://github.com/hardikpandya/stop-slop
  2. 2

    Run it

    • Clone the repository or download the folder
    • In Claude Code, add the stop-slop folder as a skill
    • Ask the model to draft or revise a piece of text and compare the output with and without the skill
  3. 3

    Get the most out of it

    • Load the reference files (phrases.md, structures.md, examples.md) when you want deeper guidance, since the README says they load on demand
    • Use the five-dimension scoring table (Directness, Rhythm, Trust, Authenticity, Density) as a self-check on drafts
    • Review examples.md for before-and-after transformations to calibrate the rules

Downloads

Source code (.zip)main branch · straight from GitHub

Strengths

  • Simple to adopt: the README lists several setup routes, including Claude Code, Claude Projects, custom instructions, and API system prompts
  • Rules are organized into phrases, structures, and examples, which makes them easy to review and edit
  • Released under the MIT license, so it can be used and modified freely

Trade-offs

  • The README does not show measured results, so how much it improves writing is unverified
  • The scoring table has 5 dimensions rated 1-10, with a revise threshold below 35/50, but the README does not explain how scores are produced
  • Rules such as banning all adverbs and em dashes are strict and may not suit every writing style

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.