natural-writing_skill

Overview

This skill helps you write like a skilled human, not a language model. It removes formulaic patterns, hollow emphasis, and overused AI vocabulary so your prose, docs, comments, and messages read natural and specific. Use it when clarity and authentic tone matter.

How this skill works

The skill scans text for common AI patterns: repeated red-flag words, filler transitions, triplet constructions, negative parallelisms, vague significance claims, and false ranges. It suggests concrete replacements, cuts hollow emphasis, and enforces style defaults like straight quotes, limited exclamation use, and natural sentence variation. It can be applied to any written output to restore directness and specificity.

When to use it

• Drafting documentation, READMEs, and API descriptions
• Writing prose, blog posts, or marketing copy that must sound human
• Editing comments in code, PR messages, and commit notes
• Polishing internal notes, emails, and policy text for clarity
• Reviewing outputs from other LLMs to remove machiney patterns

Best practices

• Prefer concrete facts over vague claims; state numbers, names, or steps when possible
• Avoid clusters of red-flag words; one occasional term is fine, many in a paragraph is a tell
• Replace formulaic constructions (’not only… but also’, lists of three) with plain, direct sentences
• Keep significance statements specific: explain why something matters with evidence or examples
• Repeat words when they refer to the same thing to preserve clarity; don’t swap synonyms for variety
• Limit punctuation tricks: no em dashes, straight quotes, and at most one exclamation per document

Example use cases

• Turn a bland feature list into concrete, actionable documentation with examples
• Rewrite a draft blog post to remove promotional puffery and add specifics
• Clean up commit messages and PR descriptions to be direct and useful
• Edit auto-generated API docs to remove vague transitions and false ranges
• Polish customer-facing emails to sound conversational and human