Anti-AI-Slop Writing

What it does

Anti-AI-Slop Writing is a directive-style skill that modifies generated text to avoid common AI fingerprints. It enforces rules like varying sentence length, avoiding predictable 'rules of three', banning specified phrases, limiting punctuation (em dashes, exclamations), and preferring active voice and concrete specifics. The skill loads a banned-words list from references and applies structural and punctuation constraints across formats (tweets, emails, articles, README sections).

When to use it

Use this skill whenever the user asks the agent to write, draft, rewrite, or humanize copy where output must not read as AI-generated — e.g., cover letters, social posts, product copy, or public-facing documentation. Trigger phrases include: 'write', 'rewrite', 'make this sound human', 'not AI', 'anti-slop'.

What's included

• Scripts: none in this repo (has_scripts=false)
• References: contains a references/banned-words.md file that the directive uses to filter vocabulary
• Instructions: a procedural rule-set covering structural rules, punctuation limits, voice calibration, and a self-check checklist to validate outputs before returning them

Compatible agents

This skill is written for use with multi-agent prompt harnesses (Claude Code, Codex, Cursor, Gemini CLI and similar assistants that support skill composition and directive injection).