# No AI Slop Editing Principles: The 17 Rules for Authentic Writing in SKILL.md

> Discover 17 editing principles in SKILL.md to eliminate AI slop and ensure authentic writing. Preserve your unique voice, facts, and cadence with effective editing.

- Repository: [Peter Yang/no-ai-slop](https://github.com/petergyang/no-ai-slop)
- Tags: best-practices
- Published: 2026-09-13

---

**The No AI Slop skill defines 17 editing principles in [`skills/no-ai-slop/SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/skills/no-ai-slop/SKILL.md) (lines 26-42) that enforce minimum effective editing—removing generic AI patterns while preserving the author’s unique voice, facts, and cadence.**

The rise of large language models has flooded technical writing with "AI slop," characterized by vague abstractions, passive constructions, and filler phrases. The **petergyang/no-ai-slop** repository provides a Codex plugin that combats this through a rigorous editorial framework. At its core, [`SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/SKILL.md) specifies **editing principles** that treat the human draft as the authoritative source, allowing only surgical interventions that enhance clarity without erasing personality.

## Core Editing Principles: Voice and Minimal Intervention

The first tier of principles establishes the philosophy of **minimum effective editing**, prioritizing preservation over polish.

**Principle 1: Preserve the writer's real voice** (line 26). The editor must notice vocabulary, cadence, humor, uncertainty, and digressions, keeping traits that feel personal to ensure the draft remains recognizably the same author after editing.

**Principle 2: Make the minimum effective edit** (line 27). Only fix AI patterns, errors, repetition, and unclear passages. Strong human sentences remain untouched to ensure edits are surgical, not wholesale rewrites.

**Principle 8: Make every sentence earn its place** (line 33). Cut empty qualifiers and throat-clearing, but keep phrases like "I think," "maybe," or "to be honest" when they reflect genuine uncertainty or rhythm rather than weakness.

**Principle 16: Preserve useful edge and character** (line 41). Retain strong opinions, blunt language, humor, profanity, self-interruptions, and honest admissions when they belong to the writer, ensuring the final draft still sounds like the original author.

## Structural and Clarity-Based Editing Principles

These rules govern how to improve flow without forcing every paragraph into a rigid template.

**Principle 3: Lead with the point when the setup adds nothing** (line 28). Cut generic throat-clearing; keep personal asides only when they add context, tension, or character to improve focus and reduce filler.

**Principle 4: Front-load only when it improves clarity** (line 29). Move conclusions earlier *if* it helps the reader, but do not force every paragraph into a point-detail-background shape. This balances readability with natural flow.

**Principle 9: Untangle sentences without flattening the cadence** (line 34). Split sentences only when genuinely hard to follow; keep longer spoken sentences, fragments, and pacing if they are clear and characteristic of the author’s natural speech pattern.

**Principle 17: Keep structure unless it hurts the piece** (line 42). Retain the author’s progression and detours when they add personality, explaining any re-ordering in the *What changed* section to balance structural clarity with voice preservation.

**Principle 15: Know the job** (line 40). Understand the piece’s purpose and audience before adjusting structure or word choice to align editing decisions with the intended impact.

## Accuracy and Specificity Principles

These editing principles enforce factual fidelity and concrete detail over abstraction.

**Principle 5: Keep the user's meaning** (line 30). Never invent claims, examples, statistics, or opinions; ask for clarification when needed to guarantee factual fidelity.

**Principle 10: Be concrete and specific** (line 35). Replace abstractions with concrete details, numbers, dates, mechanisms, or examples to turn vague statements into vivid, evidence-based prose.

**Principle 11: Use the portability test** (line 36). If a sentence could move unchanged to another person, company, or product, it is likely filler; cut or replace it with subject-specific facts to remove generic, context-free phrasing.

**Principle 12: Always show, don't tell** (line 37). Let facts, actions, examples, and consequences carry emphasis; drop commentary that merely labels something as "important" or "obvious" to encourage evidence-driven storytelling.

**Principle 13: Protect the specific fact** (line 38). Do not dilute a useful detail into generic importance; keep the concrete numbers or mechanisms that form the factual core of the argument.

**Principle 14: Make verbs do the work** (line 39). Replace weak verb phrases with direct verbs (e.g., "decided" instead of "made a decision") to strengthen sentence dynamism.

## Style and Readability Principles

These rules address the surface-level patterns that indicate AI-generated text.

**Principle 6: Open it up, don't dumb it down** (line 31). Retain substance, nuance, and precision; strip only jargon, overly long sentences, abstract nouns, and tangled structure to maintain depth while improving readability.

**Principle 7: Use active voice** (line 32). Prefer constructions like "The team shipped it Tuesday" over passive voice to make prose more direct and lively.

## Practical Implementation: Code Examples

The following examples demonstrate how to apply these editing principles according to the `petergyang/no-ai-slop` source code:

### Preserving Voice While Removing Filler

Applying **Principles 1 and 2** (preserve voice, minimum edit):

```text
Before:
"The project was absolutely amazing, and we basically crushed the timeline, which was kind of unexpected."

After:
"The project was amazing, and we crushed the timeline—something we didn’t expect."

```

The edit keeps the author’s enthusiastic tone ("amazing," "crushed") while removing filler words ("absolutely," "basically," "kind of").

### Front-Loading for Impact

Applying **Principles 4 and 7** (front-load clarity, active voice):

```text
Before:
"I think the biggest win here is the new API. The new API lets us batch requests, which reduces latency dramatically."

After:
"The new API lets us batch requests, dramatically reducing latency."

```

This moves the point forward, uses active voice, and eliminates the redundant hedge "I think."

### Concrete Specificity Over Abstraction

Applying **Principles 10 and 12** (concrete details, show don't tell):

```text
Before:
"The integration improved efficiency."

After:
"The integration cut deployment time from 40 minutes to 4 minutes."

```

The vague claim is replaced with a concrete metric, letting the fact speak for itself.

### Protecting Authorial Character

Applying **Principles 16 and 8** (preserve edge, sentence earns place):

```text
Before:
"The tool is pretty solid, but honestly, it still feels a bit clunky."

After:
"The tool works well, though it still feels a bit clunky."

```

The author’s candid admission "honestly" is retained because it adds authentic self-awareness rather than qualifying filler.

## Repository Architecture and Supporting Files

The editing principles in [`SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/SKILL.md) function as the central "style-engine" for the Codex plugin. The following files support this implementation:

- **[`skills/no-ai-slop/SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/skills/no-ai-slop/SKILL.md)**: Contains the complete 17-principle specification (lines 26-42) that drives all editing behavior.
- **[`skills/no-ai-slop/eval.md`](https://github.com/petergyang/no-ai-slop/blob/main/skills/no-ai-slop/eval.md)**: Defines the evaluation checklist that edited drafts must satisfy during workflow step 5.
- **[`.codex-plugin/plugin.json`](https://github.com/petergyang/no-ai-slop/blob/main/.codex-plugin/plugin.json)**: Metadata required for Codex-plugin integration, referencing the skill name and description.
- **[`agents/openai.yaml`](https://github.com/petergyang/no-ai-slop/blob/main/agents/openai.yaml)**: Configuration for the OpenAI model that powers the skill’s responses according to the editing principles.

## Summary

- The **17 editing principles** in [`skills/no-ai-slop/SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/skills/no-ai-slop/SKILL.md) (lines 26-42) form a minimum-effective-editing philosophy that removes AI slop without destroying human voice.
- **Voice preservation** (Principles 1, 16, 17) takes precedence over grammatical perfection, protecting cadence, humor, and edge.
- **Surgical precision** (Principles 2, 8, 9) requires changing only what is necessary to fix AI patterns, errors, or unclear passages.
- **Concrete specificity** (Principles 10-14) demands facts, numbers, and active verbs over abstract portability-test failures.
- **Structural flexibility** (Principles 3, 4, 15) improves clarity without forcing homogenized templates onto diverse writing styles.

## Frequently Asked Questions

### What is the primary goal of the No AI Slop editing principles?

The primary goal is **minimum effective editing**: removing generic AI-generated patterns ("AI slop") while preserving the author’s unique voice, factual claims, and intentional structural choices. According to the source code in `petergyang/no-ai-slop`, these principles ensure the editor serves the writer rather than imposing a standardized polish.

### Where are the editing principles documented in the repository?

All 17 editing principles are enumerated in **[`skills/no-ai-slop/SKILL.md`](https://github.com/petergyang/no-ai-slop/blob/main/skills/no-ai-slop/SKILL.md)** between lines 26 and 42. This file serves as the core specification for the Codex plugin, with additional evaluation criteria located in [`eval.md`](https://github.com/petergyang/no-ai-slop/blob/main/eval.md) and model configuration in [`agents/openai.yaml`](https://github.com/petergyang/no-ai-slop/blob/main/agents/openai.yaml).

### How does the "portability test" (Principle 11) identify AI slop?

Principle 11 states that if a sentence could move unchanged to another person, company, or product, it is likely filler. AI-generated text often produces generic statements like "We are committed to excellence" that lack specific context. The portability test forces editors to replace such phrases with subject-specific facts, immediately exposing and removing machine-generated boilerplate.

### What distinguishes "opening up" from "dumbing down" a text?

According to **Principle 6** (line 31), "opening up" means retaining substance, nuance, and precision while removing only jargon, overly long sentences, abstract nouns, and tangled structure. "Dumbing down" would strip away the complexity and depth that make an argument compelling. The skill explicitly forbids sacrificing intellectual rigor for the sake of simplified readability.