avoid-ai-writing

Detects and fixes AI writing patterns ("AI-isms") that make text sound machine-generated. Covers 21 pattern categories with a 43-entry word/phrase replacement table that maps each flagged term to a specific, plainer alternative.

When to Use This Skill

• When asked to "remove AI-isms," "clean up AI writing," or "make this sound less like AI"
• After drafting content with AI and before publishing
• When editing any text that sounds like it was generated rather than written
• When auditing documentation, blog posts, marketing copy, or internal communications for AI tells

What It Detects

21 pattern categories: formatting issues (em dashes, bold overuse, emoji headers, bullet-heavy sections), sentence structure problems (hedging, hollow intensifiers, rule of three), word/phrase replacements (43 entries like leverage→use, utilize→use, robust→reliable), template phrases, transition phrases, structural issues, significance inflation, copula avoidance, synonym cycling, vague attributions, filler phrases, generic conclusions, chatbot artifacts, notability name-dropping, superficial -ing analyses, promotional language, formulaic challenges, false ranges, inline-header lists, title case headings, and cutoff disclaimers.

Example

Prompt:

Audit this for AI writing patterns:

"In today's rapidly evolving AI landscape, developers are embarking on a pivotal journey to leverage cutting-edge tools that streamline their workflows. Moreover, these robust solutions serve as a testament to the industry's commitment to fostering seamless experiences."

Output: The skill returns four sections:

1. Issues found — every AI-ism quoted (landscape, embarking, pivotal, leverage, cutting-edge, streamline, robust, serves as, testament to, fostering, seamless, Moreover, In today's rapidly evolving...)
2. Rewritten version — "Developers are starting to use newer AI tools to simplify their work. These tools are reliable, and they're making development less painful."
3. What changed — summary of edits
4. Second-pass audit — re-reads the rewrite to catch any surviving tells

Limitations

• Does not detect AI-generated code, only prose
• Pattern matching is guideline-based, not absolute — some flagged words are fine in context
• The replacement table suggests alternatives but the best choice depends on context
• Cannot verify factual claims or find real citations to replace vague attributions