Skip to content
Get packs

Guide · Open source

How to use stop-slop: install and apply the AI-writing skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

Stop Slop is a skill file, a set of plain-text instructions, that teaches an AI model such as Claude to spot and remove predictable patterns in writing, including stock phrases, stiff sentence structures and repetitive rhythms. It is aimed at people who draft text with an LLM and want the output to read less formulaic, and at developers who want to load a writing style guide into a prompt or project. This guide shows how to get the repository, where each file fits, and how to apply the rules when you edit.

Stop Slophardikpandya/stop-slop · ★ 18.5k · MIT

Quick start

  1. 1Clone the repository. Run the git clone command from the README to download stop-slop, including SKILL.md and the references folder.
  2. 2Pick your setup route. Choose Claude Code, Claude Projects, custom instructions or API system prompts, depending on how you use the model.
  3. 3Load the core rules. Add SKILL.md as a skill, upload it to project knowledge, copy its rules into custom instructions, or include it in the system prompt.
  4. 4Provide the reference files. In Claude Projects, upload the reference files too. The README says reference files load on demand when you use the API.
  5. 5Draft and ask for a revision. Ask the model to write or edit your text under the skill's rules, then check the output against the banned phrases and structures.
  6. 6Score the result. Rate the draft on the five dimensions. If the total is below 35 out of 50, revise it.

What stop-slop is and what it is not

Stop Slop is not an app, an editor or a detector. The README describes it as a core SKILL.md file plus reference files. Those files list phrases to remove, structures to avoid and before-and-after examples, and they tell a model how to rewrite text.

The project is about style rules. It does not claim to identify AI-generated text, so do not rely on it to prove where a piece of writing came from.

  • Core instructions: SKILL.md
  • Phrases to remove: references/phrases.md
  • Structural patterns to avoid: references/structures.md
  • Before/after transformations: references/examples.md

Get the stop-slop files

The README lists a single install command, which clones the repository. No prerequisites are listed. Run the command in the folder where you want the skill to live.

The README shows the repository layout, with SKILL.md at the top level and the three reference files in a references folder.

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

Use stop-slop in Claude Code, Claude Projects or custom instructions

The README gives four routes depending on how you work with Claude. In Claude Code, add the stop-slop folder as a skill. In Claude Projects, upload SKILL.md and the reference files to project knowledge. If you use custom instructions, copy the core rules from SKILL.md into them.

For API calls, include SKILL.md in your system prompt. The README says the reference files load on demand.

  • Claude Code: add this folder as a skill
  • Claude Projects: upload SKILL.md and reference files to project knowledge
  • Custom instructions: copy core rules from SKILL.md
  • API calls: include SKILL.md in your system prompt; reference files load on demand

Know what the rules catch

The README groups the rules into banned phrases, structural clichés and sentence-level rules. Banned phrases cover throat-clearing openers, emphasis crutches, business jargon, all adverbs, vague declaratives and meta-commentary. Structural clichés cover binary contrasts, negative listings, dramatic fragmentation, rhetorical setups, false agency, a distant narrator voice and passive voice.

Sentence-level rules include no Wh- sentence starters, no em dashes, no staccato fragmentation, no lazy extremes and required active voice.

  • Banned phrases: openers, emphasis crutches, jargon, adverbs, vague declaratives, meta-commentary
  • Structural clichés: binary contrasts, negative listings, dramatic fragmentation, rhetorical setups, passive voice
  • Sentence rules: no Wh- starters, no em dashes, active voice

Score drafts with the five-dimension check

The skill includes a scoring table. You rate a draft from 1 to 10 on five dimensions: Directness (statements or announcements), Rhythm (varied or metronomic), Trust (respects reader intelligence), Authenticity (sounds human) and Density (anything cuttable).

The README says that a total below 35 out of 50 means the draft should be revised.

  • Directness: statements or announcements?
  • Rhythm: varied or metronomic?
  • Trust: respects reader intelligence?
  • Authenticity: sounds human?
  • Density: anything cuttable?

Adapt the rules for your own publication

The rules are general. The README does not explain how to change them for a particular publication's style guide, so if you need house style, you will need to edit the files yourself and test the results.

What people use it for

  • Editing an AI-drafted blog post to remove stock phrases and emphasis crutches
  • Creating a checklist of banned phrases and structural habits for a human editing pass
  • Loading a writing style guide into a Claude Project so every draft follows the same rules
  • Adding SKILL.md to an API system prompt for a text-generation workflow

Frequently asked questions

Is stop-slop a standalone app?

No. The README describes only instruction files: a core SKILL.md and reference files. It does not describe its own interface, so you use it through Claude, another LLM, or by reading the files yourself.

How do I install stop-slop?

The README gives one install command: git clone https://github.com/hardikpandya/stop-slop. No prerequisites are listed. After cloning, you add the files to Claude Code, Claude Projects, custom instructions or a system prompt.

Is stop-slop free and what is its licence?

It is released under the MIT licence. The README says: use freely, share widely.

Can stop-slop detect AI-generated text?

No. The project is about style rules, not detection. Do not use it to guarantee that text is or is not AI-written.

Does it work with LLMs other than Claude?

The README says the skill teaches Claude or any LLM to catch and remove AI patterns, so you can load the same files into another model that accepts instructions. Check the output yourself.

Are there alternatives or comparable projects?

The analysis for this directory lists no comparable projects. Any other style guide you load into a prompt would be a different project, and the README does not compare them.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Stop Slop suits people who draft with Claude or another LLM and want a concrete checklist of phrases and structures to remove, and developers who want to load that checklist into a system prompt or project. It is a set of instruction files, not an app, a detector, or a publication-specific style guide, so skip it if you need any of those.

Install options & downloads