No AI Slop Skill Modes Explained: Edit, Detect, and Generate

The No AI Slop skill operates in three distinct modes—Edit, Detect, and Generate—that are automatically selected based on your command trigger phrase to either clean AI slop from text, identify slop patterns, or create satirical AI-generated content.

The petergyang/no-ai-slop repository provides a ChatGPT/Codex plugin skill designed to identify and eliminate "AI slop," which refers to formulaic, generic text patterns produced by large language models. Understanding the different No AI Slop skill modes allows you to sanitize existing writing, audit text for slop patterns, or generate deliberately cringe-worthy examples for testing and satire.

How Mode Selection Works

The skill parses your command line input to determine which operational path to follow, as documented in the "How to use No AI Slop" section of README.md (lines 31-55). Each mode applies a specific rule set defined in the skill's configuration files to process your request appropriately.

Edit Mode

Trigger phrase: /no-ai-slop (your writing)

Edit mode removes AI-slop patterns from supplied text while preserving your personal voice. The skill applies editing rules defined in SKILL.md to strip generic phrases and reports the specific changes made to your content.

/no-ai-slop I think we should consider the best approach because it's not X. It's Y.

Detect Mode

Trigger phrase: /no-ai-slop is this slop? (your writing)

Detect mode scans input for known slop patterns and returns each match quoted directly, without attempting to guess the author or rewrite the content. This audit function references the pattern definitions stored in the skill's configuration.

/no-ai-slop is this slop? The future isn't coming. It's already here.

Generate Mode

Trigger phrase: Draft an AI slop post about (topic) (used after the /no-ai-slop prefix)

Generate mode produces deliberately "cringe" AI-slop text on the given topic. Unlike the other modes, this creates rather than removes AI slop, making it useful for satire or testing detection capabilities on known-bad examples.

/no-ai-slop Draft an AI slop post about remote work trends

Technical Implementation and Configuration Files

The No AI Slop skill relies on several key files stored in the repository to execute these modes:

• skills/no-ai-slop/SKILL.md – Contains the editing rules and overall workflow that govern how Edit mode sanitizes text and how the skill processes commands.
• skills/no-ai-slop/eval.md – Defines the checks the skill runs on its output to ensure quality and accuracy across all three modes.
• .codex-plugin/plugin.json – Stores the metadata used for ChatGPT/Codex plugin integration, enabling the skill to parse commands and route them to the appropriate mode handler.
• scripts/build_plugin.py – Builds and validates the plugin package, ensuring that mode selection logic and rule sets are properly packaged for deployment.

Summary

• The No AI Slop skill provides three modes: Edit (clean text), Detect (audit text), and Generate (create satirical slop).
• Mode selection is automatic based on trigger phrases parsed from the command line, as implemented in the repository's core logic.
• Edit mode sanitizes content using rules from SKILL.md, while Detect mode identifies patterns without rewriting.
• Generate mode creates intentionally generic AI text for testing or humorous purposes.
• Configuration files including eval.md and .codex-plugin/plugin.json support the technical implementation across all modes.

Frequently Asked Questions

How does the No AI Slop skill decide which mode to use?

The skill analyzes your command input for specific trigger phrases defined in the source code. If you prefix with /no-ai-slop followed immediately by text, it enters Edit mode. If you include "is this slop?" after the prefix, it switches to Detect mode. If you ask it to "Draft an AI slop post," it activates Generate mode to produce satirical content.

Can I use Detect mode on my own writing to check for accidental AI slop?

Yes. According to the petergyang/no-ai-slop source code, Detect mode scans any supplied text for known slop patterns and returns quoted matches without making assumptions about authorship. This makes it ideal for auditing your own content before publication to ensure it maintains an authentic voice.

What file contains the rules for identifying AI slop patterns?

The editing rules and workflow definitions are stored in SKILL.md within the skill directory. The eval.md file contains additional validation checks that ensure output quality across all three operational modes, while the detection patterns are referenced from the skill's core configuration.

Is Generate mode actually useful, or is it just for entertainment?

Generate mode serves both educational and testing purposes beyond pure entertainment. It produces deliberately formulaic AI text that developers can use to test the Detect mode's pattern matching accuracy or to educate teams about what constitutes generic AI slop versus authentic human writing.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}