de-ai-ify

What This Skill Does

The de-ai-ify skill acts as a sophisticated linguistic filter designed to strip away the tell-tale artifacts of Large Language Model generation. Unlike simple prompts that ask an AI to "sound human," this tool utilizes a systematic detection engine trained on over 1,000 document pairs. It targets 47 specific stylistic indicators, including overused transitions like "moreover," vacuous cliches such as "fast-paced world," and robotic sentence structures. By normalizing readability and injecting varied sentence rhythms, it restores an authentic voice to your drafts. Whether you are polishing corporate reports, blog posts, or email communication, this skill ensures your writing feels grounded, intentional, and personal rather than algorithmically generated.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/brianrwagner/brw-de-ai-ify

Ensure you have the latest version of the OpenClaw agent installed before proceeding to ensure compatibility with the scoring engine.

Use Cases

• Content Marketing: Quickly sanitize AI-drafted blog posts so they sound like expert-authored insights.
• Corporate Communications: Rewrite internal memos to remove "corporate-speak" that creates distance between leadership and employees.
• Technical Documentation: Convert stiff, robot-like tutorials into clear, actionable, and conversational guides.
• Ghostwriting: Use the skill to iterate through draft versions of copy, ensuring the final output retains your unique style while adhering to professional quality standards.

Example Prompts

1. "@openclaw, please de-ai-ify ./drafts/marketing_copy.txt and give me a summary of the changes made."
2. "@openclaw de-ai-ify ./reports/q3_update.md --score-threshold 9"
3. "@openclaw, take the text in ./blogs/trends.txt, run the de-ai-ify skill, and output the result to ./blogs/trends_polished.txt"

Tips & Limitations

While the skill is excellent at identifying patterns, it is a tool for refinement, not factual correction. Always review the "Change tracking" report provided after execution to ensure no unintended meaning was lost during the structural optimization. For documents with highly technical or proprietary jargon, you may occasionally need to add specific terms back into the output, as the "Corporate Buzzword" filter can be aggressive in its pursuit of clarity. Use the --score-threshold flag to control how strictly the skill modifies your work; a setting of 8 is recommended for most business scenarios, while 9-10 is reserved for high-stakes, professional creative writing.