How to Use No AI Slop in Detect Mode: Identify AI Writing Patterns

No AI Slop's Detect mode scans text for AI-generated patterns like binary contrasts and throat-clearing openers, listing each violation with line quotes and fix suggestions without rewriting your draft.

No AI Slop is a Codex-style skill available in the petergyang/no-ai-slop repository that helps writers identify and eliminate generic AI-generated language. When operating in Detect mode, the skill acts as a diagnostic tool rather than an editor, analyzing your draft against 20+ regex-based patterns to flag specific AI-slop constructions. This mode preserves your original voice while providing actionable feedback on exactly which lines need attention and how to fix them.

What Is No AI Slop Detect Mode?

Detect mode is one of two primary job types defined in SKILL.md, alongside the Edit mode. According to the skill definition, Detect mode identifies AI-slop patterns, quotes the offending lines verbatim, and suggests brief fixes—without rewriting the draft or guessing whether an AI actually wrote the content.

As specified in eval.md, the detection workflow must output a structured report that names each pattern found, includes the exact quoted line, and provides a concise fix description. This ensures the user maintains full control over the editing process while receiving precise guidance on eliminating robotic phrasing.

The Detect Mode Workflow

When you invoke the skill with a detection request, the system follows a four-step pipeline defined in the skill metadata and evaluation checklist.

Step 1: Job Identification

The skill parses your prompt to determine which branch to execute. If the prompt contains the phrase "is this slop?", the skill automatically selects the Detect job rather than the Edit job. This trigger phrase is explicitly defined in the "Two jobs" section of SKILL.md.

Step 2: Pattern Matching

Once in Detect mode, the skill iterates through a comprehensive list of 20+ regex-based patterns enumerated in the "Patterns to cut" subsection of SKILL.md. These patterns target specific AI-writing hallmarks including:

• Binary contrasts (e.g., "It's not X. It's Y.")
• Throat-clearing openers (e.g., "Here's the thing...")
• Faux-insight setups and colon reveals
• Fake-profound kickers

The matcher records every instance where these patterns appear in your text.

Step 3: Report Generation

For each matched pattern, the skill generates a report entry following the strict format mandated by eval.md. Each entry contains:

• The pattern name (e.g., "Binary contrast")
• The original line quoted verbatim
• A short fix description (e.g., "Replace with a direct statement")

The evaluation checklist explicitly requires that Detect responses do not rewrite the draft but instead provide these specific diagnostic elements.

Step 4: Follow-up Editing

After delivering the detection report, the skill prompts you to decide whether to proceed with editing. This preserves the original voice while offering the option to apply fixes manually or switch to Edit mode for automatic rewriting.

Practical Examples of No AI Slop Detect Mode

You can invoke Detect mode through chat interfaces or command-line installations.

Chat Interface Usage

In any Codex or ChatGPT interface supporting the skill, trigger detection with the specific trigger phrase:

/no-ai-slop is this slop?
It's not a bug. It's a feature.
Here's the thing, the code works.

Result:

• Binary contrast – "It's not a bug. It's a feature." → Replace with a direct statement, e.g., "The feature adds functionality."
• Throat-clearing opener – "Here's the thing, the code works." → Remove the opener and start with the point.

Command Line Installation and Usage

First, install the skill globally using the command-line installer referenced in README.md:

npx skills add petergyang/no-ai-slop --skill no-ai-slop --global --yes

Then invoke detection in your terminal:

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

Result:

• Fake-profound kicker – "The future isn't coming. It's already here." → Delete the line; the draft ends on a concrete statement.

Key Configuration Files Behind Detect Mode

Three files govern how Detect mode operates:

• SKILL.md – Defines the Detect job workflow, lists the "Patterns to cut," and specifies the trigger phrase "is this slop?"
• eval.md – Mandates the reporting format requiring pattern names, quoted lines, and short fixes without draft rewriting
• plugin.json – Declares "Detect" as a supported capability in the Codex/ChatGPT plugin manifest, exposing the detection-only operation to the interface

Summary

• No AI Slop Detect mode identifies AI-generated patterns without automatically rewriting your text.
• Trigger detection by including the phrase "is this slop?" in your prompt.
• The skill scans for 20+ specific patterns including binary contrasts, throat-clearing phrases, and fake-profound kickers.
• Output format requires pattern names, verbatim line quotes, and concise fix suggestions as defined in eval.md.
• Installation is available via npx skills add petergyang/no-ai-slop for command-line usage.

Frequently Asked Questions

What specific patterns does No AI Slop Detect mode identify?

According to SKILL.md, Detect mode targets over 20 regex-based patterns including binary contrasts ("It's not X. It's Y."), throat-clearing openers ("Here's the thing"), faux-insight setups, colon reveals, and fake-profound kickers. The complete enumeration appears in the "Patterns to cut" subsection of the skill definition.

Does Detect mode automatically rewrite my text?

No. As specified in eval.md, Detect mode only identifies patterns, quotes the offending lines, and suggests fixes. It does not rewrite the draft or guess whether an AI wrote the content. You must explicitly request editing in a follow-up prompt if you want automatic rewriting.

How do I install No AI Slop for command line use?

Install the skill globally using the Node Package Executor command: npx skills add petergyang/no-ai-slop --skill no-ai-slop --global --yes. After installation, you can invoke /no-ai-slop is this slop? directly in your terminal to analyze text.

What's the difference between Detect mode and Edit mode?

Detect mode diagnoses problems by listing matched patterns with quoted lines and fix suggestions, preserving your original draft. Edit mode actively rewrites the content to eliminate AI-slop patterns while maintaining your voice. The job branch is determined by whether your prompt contains "is this slop?" (Detect) or requests direct editing (Edit).

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