Guide · Agent skill
How to use the Anti-Slop Copywriting agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The antislop-copywriting skill is a set of writing rules that an AI coding agent loads when it writes or edits marketing copy, product copy, headlines, calls to action or other prose. It checks drafts for patterns that make text sound machine-made, such as buzzwords like 'unlock' and 'elevate', unsupported claims like 'trusted by thousands', and chatbot sign-offs. This guide explains what the skill does, how to install it, how to trigger it with plain requests, and what it needs to work.
Anti-Slop Copywritingmiqdadbadjuber/anti-slop · MIT Download the skill Quick start
- 1Get the skill file. Download the antislop-copywriting ZIP from this page.
- 2Unzip it into the right folder. Unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the skill.
- 4Add the antislop.md core file. Obtain the companion antislop.md file, since the skill is meant to be read together with it.
- 5Ask for copy work in plain language. Request a headline, CTA, landing-page rewrite or prose edit and include the real facts you want kept.
- 6Review the output. Check every claim and name the agent kept, since the skill removes style patterns but does not verify facts.
What does the antislop-copywriting skill make the agent do?
Once the skill is loaded, the agent checks its own writing against a list of named patterns. Each pattern has a short explanation of why it reads as AI, a bad example (the 'before') and a plainer rewrite (the 'after'). The patterns covered in the reviewed file include empty AI vocabulary, significance inflation, unsupported trust claims, weasel attributions, chatbot closers, fake-candid openers, signposting phrases, all-caps emphasis and actorless passive voice.
When rewriting, the skill tells the agent not to add facts, names, numbers, dates, quotes or citations that are not already in the source text or supplied by you. If a sentence needs real detail to work, the skill says to ask for it or write the plainer version without it.
- Replaces buzzwords such as 'unlock', 'elevate' and 'seamless' with concrete wording.
- Removes claims like 'trusted by thousands of teams' unless a real customer or number can be named.
- Strips chatbot filler such as 'I hope this helps!' from text meant for readers.
- Keeps your voice: the skill warns against making copy so bland that it reads as machine-made.
How do you install the skill?
You can install the skill in two ways. For Claude Code, download the ZIP from the skill's page on this directory and unzip it into your user skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it to apply only to one project. For the Claude apps, you can upload the skill instead under Settings, then Capabilities, then Skills.
The skill is a single file, SKILL.md, with a short header that names the skill and describes when to use it. Keep the file in its own skill folder when you unzip it so the agent can find it.
- Option 1: download the ZIP from this page and unzip it into ~/.claude/skills/ for all your projects.
- Option 2: unzip it into .claude/skills/ inside one project folder.
- Option 3: in the Claude apps, go to Settings, then Capabilities, then Skills, and upload it.
Do you also need the antislop.md core file?
Yes. The skill is written to be read together with a core rules file called antislop.md. That file holds the general mechanism and the hard bans, and this skill adds copy-specific detail on top. The skill refers to core rules by number, such as R-16 and R-36.
The core file is not included in the files reviewed for this page. Without it, the skill's rule references cannot be checked, so you will need to obtain antislop.md separately before relying on the skill for copy work.
- The skill header tells the agent to load it with the core.
- Rule numbers cited in the skill, such as R-17 or C-5, come from the core file and cannot be verified from the skill alone.
How do you trigger the skill with natural requests?
You do not need special commands. The skill is meant to load when the task involves writing or editing prose: headlines, tone, calls to action, landing-page text, product copy or anti-AI-writing patterns. Ask for that kind of work in plain language and mention the copy you want reviewed.
If you want to be sure the rules are applied, say so in the request, for example by naming the skill or asking for the core and copy rules to be followed.
- 'Rewrite this landing page hero so it has no buzzwords and no unsupported claims.'
- 'Edit this product description and remove any chatbot sign-off or filler.'
- 'Review these three CTAs for generic AI wording and suggest plainer versions.'
- 'Use the antislop copywriting rules on this email draft. Do not add any statistics.'
What do before and after examples look like?
The skill shows each pattern as a pair. For example, a sentence like 'Unlock the power of seamless collaboration to elevate your team's journey' becomes 'Work with your team in one shared space.' The rewrite is shorter and states only what the product does.
Another example: 'Trusted by thousands of teams worldwide' is replaced with a line naming real customers, and if there are no real customers to name, the skill says to cut the claim entirely. The skill never swaps in a made-up number to make the sentence sound stronger.
- Before: 'Experts say this approach dramatically improves conversion.' After: name the real source, or cut the sentence.
- Before: 'Honestly? It depends on how often you'll use it.' After: 'Whether it is worth the price depends on how often you'll use it.'
- Before: 'Let's dive into how caching works. Here's what you need to know.' After: the first fact, stated directly.
What are the safety notes for scripts and network access?
According to the reviewed details, the skill runs no scripts and uses no network access. It consists of instructions in SKILL.md. The skill header lists the tools it is allowed to use as Read, Write, Edit, Glob and Grep, which are for reading and changing files in your project.
Because the agent can write and edit files, review its changes before you publish copy, as you would with any text an agent edits.
- Runs scripts: no.
- Uses network: no.
- Allowed tools: Read, Write, Edit, Glob, Grep.
- The skill does not check facts. It removes style patterns and adds no sourcing, so verify claims yourself.
What tips help you get better copy?
Give the agent the real details first: product names, real customers, real numbers and your brand voice. The skill can only remove vague wording; it cannot supply specifics. Keep a voice in the request if you have one, because the skill itself warns against copy with no personality.
For longer work, ask the agent to run the copy checklist described at the end of the skill file alongside the core Delivery Gate, and to list each change it made.
- Paste the real facts into the request so the rewrite has something specific to say.
- Do not ask for invented statistics or testimonials; the skill forbids them.
- For factual or research writing, check every claim separately.
- Read the output aloud to catch remaining stiff or generic lines.
What people use it for
- Tightening a landing page hero and calls to action so they say what the product does.
- Cleaning up product descriptions drafted by an AI so no chatbot filler reaches customers.
- Editing an editor's draft for telltale phrasing such as signposting and fake-candid openers.
- Rewriting value propositions with a plainer, specific claim the team can back up.
Frequently asked questions
How do I install the antislop copywriting skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings, then Capabilities, then Skills.
Is the antislop copywriting skill free?
The reviewed listing states the licence is MIT. Check the licence file in the source repository for the full terms before you reuse or redistribute it.
What do I need to run the skill?
You need a SKILL.md-compatible agent runtime and the companion antislop.md core file. The core file is referenced by the skill but was not included in the reviewed files, so you must get it separately.
Does the skill check whether my facts are true?
No. The skill only removes style patterns and adds no sourcing. It can leave unverified statements in place, so you need to verify factual claims yourself.
Does the skill run scripts or use the internet?
According to the reviewed details, it runs no scripts and uses no network access. It is made up of instructions in SKILL.md.
Why does the skill still reference rule numbers I cannot see?
The skill cites core rules such as R-17 and C-5 that live in antislop.md. Without that file, those rule numbers cannot be checked from the skill alone.
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
This skill suits marketers, founders and editors who want headlines, landing pages and calls to action with less generic, AI-sounding wording and no invented claims. It is not a fact-checker, so pick a different approach for research or any writing where every claim must be verified, and make sure you have the antislop.md core file before relying on it.
Download the skill