No AI Slop: evidence-led editing, packaging and evaluation
Choose No AI Slop, a style guide or manual review by the editing task
Compare explicit pattern findings with general style guidance and editorial judgment, without confusing pattern detection with authorship detection.
What you will learn
- Start with the output you want to inspect
- Do not substitute an authorship accusation for editorial evidence
- Adopt only for a bounded use case
Before you start
- A draft whose facts can be checked
- Basic understanding of assistant instructions and plugin scope
Inspect pattern evidence, preserve meaning and distinguish package checks from unmeasured editing outcomes.
Key takeaways
- Pattern findings, house style and human judgment solve different parts of editing.
- A pattern is not proof of authorship.
- Detection-only mode can preserve author control.
Start with the output you want to inspect
If you need a short list of problematic lines with named patterns, the skill’s detection contract provides a specific output shape. If you need consistent terminology across a publication, a maintained style guide may be the more important input. If a draft contains delicate argument or authorial voice, an editor must still judge the proposed changes in context.
These approaches can be combined. A style guide specifies house preferences, the skill identifies recurring filler, and a reviewer protects meaning. Conflicting instructions need an explicit priority: an approved technical term or exact quotation should not be erased simply because it resembles a prohibited word or construction.
Do not substitute an authorship accusation for editorial evidence
An authorship detector tries to infer how text was produced. No AI Slop’s documented detection mode instead cites visible writing patterns and forbids that inference. A human can use a cliché and a model can write a sentence without one. Findings should support an editing decision, not an allegation about the writer.
A simple word list can locate exact strings cheaply, but it cannot by itself decide whether a word is technical, quoted or part of a recognizable voice. The skill asks a model to use context, introducing judgment and variability. Neither method should be treated as an automatic factual acceptance gate for a public article.
Adopt only for a bounded use case
Trial the skill on the content type that currently causes editing friction. Define rejection conditions such as changing protected numbers, dropping uncertainty or rewriting during detection. Compare the time needed to inspect and repair results against your current process. You may choose findings-only mode when the author wants complete control over revisions.
Do not choose it solely because a sample sounds less like a stereotype of AI writing. The decision should depend on meaning retained, evidence available and the burden placed on the reviewer. No competing product benchmark or universal superiority claim is supported by the source inspection in this series.
Decision guide
| Criterion | Option A | Option B |
|---|---|---|
| Best when | You need predictable behavior and easy auditing | You need adaptive optimization and have reliable telemetry |
| Main risk | May leave performance on the table | Can become difficult to explain or debug |
Implementation steps
- 1
Define the required output and protected content.
- 2
Resolve conflicts with the house style guide.
- 3
Test representative drafts and rejection conditions.
- 4
Compare review burden before adopting.
Copy-ready example
{
"preferredMode": "detect",
"protectedContent": [
"exact quotes",
"technical terms",
"numbers"
],
"rejectIf": [
"rewrites without permission",
"guesses authorship",
"adds evidence"
],
"comparisonMeasured": false
}Frequently asked questions
Should I replace editorial review with this skill?
No. Context, meaning and the author’s intent still require review.
Is a word-list checker equivalent?
It can match strings, but contextual exceptions and preservation decisions need additional judgment.
Sources
- No AI Slop / skills/no-ai-slop/SKILL.mdSource checked 2026-09-14
- No AI Slop / skills/no-ai-slop/eval.mdSource checked 2026-09-14
- No AI Slop / README.mdSource checked 2026-09-14