natural-writinglisted
Remove AI writing patterns from any prose a person will read, and keep them out of prose you write. EDIT AND AUDIT TRIGGERS: "deslop", "de-AI", "humanize", "make it sound human", "clean up this writing", "find the AI tells", "sounds robotic", "reads like a chatbot", "sounds like AI wrote it", "full of buzzwords". WRITE TRIGGERS: "write", "draft", "write up", "put together", "turn these notes into", or naming where it goes -- to send, to post, to publish, to share with someone. ARTIFACTS: any email, any letter, blog post, newsletter, social post, announcement, press release, any report, meeting notes, any proposal, README, docs, release notes, website or marketing copy, slide text, UI microcopy, artifact or published page. NOT TRIGGERED BY: "explain", "what is", "how do I", or any answer that stays in the conversation rather than going to a reader. NOT FOR: code, comments, commit messages, or interface typesetting such as quotes, dashes and ellipses as rendered.
LarrySequino/skills · ★ 1 · AI & Automation · score 72
Install: claude install-skill LarrySequino/skills
# Natural Writing: Prose Without AI Patterns
Strip predictable AI patterns from writing. Make prose sound like a specific human wrote it, not like a language model generated it.
## When rules conflict
The rules below do not carry equal weight, and they do collide. When two of them point opposite ways, the higher item wins:
1. The user's explicit instructions, and what the destination requires.
2. Truth: facts, relationships, quotations, qualifiers, causality, scope, commitments.
3. The author's voice as supplied, and their deliberate choices.
4. Genre and audience convention.
5. Clarity, density, and the quality of the argument.
6. The anti-slop pattern heuristics in the core rules.
7. Detector-facing signals and house style.
A lower item never overrides a higher one. Where a rule below says it defers, this is the order it defers to, and the numbers refer to this list.
## What a pattern match is worth
The patterns here are statistically more common in LLM output, but humans on autopilot produce the same shapes, whether under deadline, in an unfamiliar genre, or in a second language. A checklist is single-feature reasoning and the detectors that work are not, which is the durable reason a hit here proves nothing about who wrote it (`references/preflight.md`). So: these patterns are worth fixing in any prose, but never treat them as proof of AI authorship for a consequential decision (academic integrity, hiring, attribution). When auditing someone else's text, report pat