How to use the Stop Slop agent skill: install and trigger it

What is an agent skill, and what does Stop Slop make the agent do?

An agent skill is a folder of instructions that an AI agent can load when a task calls for it. Stop Slop's main file, SKILL.md, tells the agent to apply a set of writing rules when it drafts, edits or reviews prose.

The rules cover filler phrases, formulaic structures, passive voice, vague declaratives and quotable lines. The agent also scores text from 1 to 10 on directness, rhythm, trust, authenticity and density, and revises anything scoring below 35 out of 50.

• Cut throat-clearing openers, emphasis crutches and adverbs.
• Replace passive voice with a human subject doing something.
• Remove binary contrasts such as 'not X, it's Y' and dramatic fragments.
• Vary sentence length and remove em dashes.
• Name specifics instead of writing vague declaratives.

How do you install the Stop Slop skill?

Download the stop-slop ZIP from this page and unzip it. Then put the stop-slop folder in one of these locations, depending on how you work.

For a personal setup available across projects, place it in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder. You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/stop-slop/
• Project: .claude/skills/stop-slop/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the skill

How do you trigger the skill with natural requests?

You do not need a special command. The skill's description says it applies when drafting, editing or reviewing text, so ask for that work in plain language and name the skill if you want to be sure it runs.

The metadata trigger for the skill is 'Writing prose, editing drafts, reviewing content for AI patterns'. Requests that match that description are the ones it is designed for.

• 'Edit this newsletter draft with the stop-slop skill.'
• 'Review this essay for AI writing patterns.'
• 'Draft a blog intro about our product launch and apply stop-slop.'
• 'Tighten this paragraph and remove filler phrases.'

What do the reference files add?

SKILL.md points to three reference files. references/phrases.md lists phrases to remove, references/structures.md covers structural patterns to avoid, and references/examples.md contains before-and-after transformations.

The README describes these files as loading on demand in API setups. The CHANGELOG lists example phrases such as 'Here's what I find interesting' and 'Not X. But Y.' that the reference files target.

What does the skill need, and is it safe to run?

The skill lists no requirements. It does not run scripts and does not use the network. The files included are CHANGELOG.md, README.md and SKILL.md, with the reference folder described in the README.

The skill is MIT licensed and its author is Hardik Pandya. Because it only gives the agent writing rules, it has no install scripts to review, but you should still read SKILL.md before installing any skill you did not write.

Tips for getting good results

Run the skill on a draft you have written, then read every change. The rules are strict, and some edits can alter meaning or remove tone you wanted to keep.

Give the agent the context it needs, such as the audience and the publication, so it does not strip a voice you chose on purpose.

• Check that quoted material and names are not changed.
• Keep a copy of your original draft to compare.
• Use it as a checklist for your own edits, not a final judge of quality.