How to use the No AI Slop Editor agent skill

What is an agent skill?

An agent skill is a folder of instructions that an AI agent loads when a task matches the skill's description. The skill tells the agent how to behave for that kind of task. Nothing here is a separate app: you keep working in your agent, and the skill changes how it handles drafts.

What the no-ai-slop skill makes the agent do

The skill has two jobs. By default it edits: it makes the minimum effective edit, cuts stock AI phrasing, filler, vague claims and empty qualifiers, keeps your voice and opinions, and returns the revised draft with a What changed section explaining the edits.

The second job is detection. If you only want a check, the agent names each AI-style pattern it finds, quotes the line, and gives a short fix. It does not rewrite the draft in that mode, and it does not score the draft or guess whether AI wrote it.

• Edit mode: revised draft plus a What changed list
• Detect mode: each pattern, its quoted line and a quick fix, no rewrite
• The skill lists banned words such as delve, leverage and robust, and common patterns such as binary contrasts and colon reveals

Install the skill from the ZIP

Download the no-ai-slop ZIP from this page and unzip it. The skill folder contains agents/openai.yaml, eval.md and SKILL.md. Put the folder in one of these locations:

• Personal skills: ~/.claude/skills/
• Project skills: .claude/skills/ inside your project folder
• Claude apps: upload the skill under Settings → Capabilities → Skills

Trigger the skill with natural requests

You do not need a special command. The skill's description tells the agent when to use it, so ask in plain language. Paste your draft and state what you want. The skill asks one question if the audience or format is unclear, and asks what the reader should think, feel or do if the goal is unclear.

• Make this draft clearer and more direct, keep my voice: [paste draft]
• Make this less AI-sounding before I publish it: [paste draft]
• Scan this draft for AI-sounding patterns, quote each one and suggest a fix, don't rewrite it: [paste draft]
• Does this paragraph read as AI? Point to the lines and don't guess about authorship.

Example: an edit and a check

Suppose a line reads: "The question isn't the model. It's the eval." In edit mode, the skill rewrites it as a direct statement, such as "The eval matters more than the model." In detect mode, it quotes the same line, labels it a binary contrast, and suggests stating the point directly without rewriting the rest of the draft.

The skill also keeps what makes your writing yours. Strong opinions, blunt language, humor and honest admissions are meant to stay.

What it needs and safety notes

The listing says the skill requires nothing beyond the agent itself, does not run scripts, and does not use the network. Because it runs no scripts and makes no network calls, the only thing it changes is the text the agent returns to you.

The skill cannot verify facts. It keeps your meaning and is designed not to invent stats, examples or quotes, but you should still check every claim before publishing.

Tips for better results

Give the agent the full draft and the audience, because the skill edits differently for a newsletter than for a technical doc. Ask for detect mode first if you want to learn the patterns yourself. Read the What changed list and reject any edit that changes your meaning.

• Use detect mode on a draft you wrote yourself to see which patterns recur in your writing
• Keep your original file so you can compare
• Run the check again after you revise, since the skill's eval checks expect a fixed draft before it is returned