Audit and Rewrite AI-Generated Text
Audits and rewrites text to remove AI-isms - 21 pattern categories and a 43-entry replacement table - so writing doesn't sound machine-generated.
Why it matters
Eliminate AI writing tells and ensure your content sounds authentically human. This skill audits text for AI patterns and provides rewritten versions that are clear, concise, and natural.
Outcomes
What it gets done
Detect and flag 21 categories of AI writing patterns.
Rewrite flagged text to sound more human and less robotic.
Provide a second-pass audit to confirm AI-isms are removed.
Identify specific word/phrase replacements for common AI-isms.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/ag-avoid-ai-writing | bash Overview
Avoid AI Writing - Audit & Rewrite
Detects and fixes AI writing patterns ('AI-isms') across 21 categories - formatting issues, hedging, template phrases, vague attributions, promotional language, and more - using a 43-entry word/phrase replacement table, then re-audits the rewrite for anything that survived. Use it after drafting content with AI and before publishing, or whenever asked to remove AI-isms or make writing sound less machine-generated. It only catches prose, not AI-generated code, and can't verify facts or find real citations.
What it does
This skill audits text for "AI-isms" - the patterns that make writing sound machine-generated - across 21 categories: formatting issues (em dashes, bold overuse, emoji headers, bullet-heavy sections), sentence-structure problems (hedging, hollow intensifiers, rule of three), template and transition phrases, 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, cutoff disclaimers, and a 43-entry word/phrase replacement table (e.g. leverage→use, utilize→use, robust→reliable) mapping flagged terms to plainer alternatives.
When to use - and when NOT to
Use it 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, or when auditing documentation, blog posts, marketing copy, or internal communications for AI tells. It's prose-only: it does not detect AI-generated code. Its pattern matching is guideline-based rather than absolute, so some flagged words are fine depending on context, and it cannot verify factual claims or supply real citations to replace vague attributions - only flag that an attribution is vague.
Inputs and outputs
Input is a passage of AI-drafted (or AI-sounding) text. Output is four sections: "Issues found" (every AI-ism quoted directly from the source, e.g. "landscape," "embarking," "pivotal," "leverage," "Moreover"), a rewritten version using plainer language, a "what changed" summary of the edits, and a second-pass audit that re-reads the rewrite to catch any surviving tells. For example, "In today's rapidly evolving AI landscape, developers are embarking on a pivotal journey to leverage cutting-edge tools..." becomes "Developers are starting to use newer AI tools to simplify their work. These tools are reliable, and they're making development less painful."
Integrations
The skill is self-contained - it doesn't call external services or APIs, relying entirely on its own pattern catalog and replacement table to detect and rewrite text.
Who it's for
Writers, editors, and teams publishing AI-assisted content who want a systematic, repeatable check for AI writing tells before publication, rather than relying on ad hoc "does this sound robotic?" judgment.
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.