humanizer-enhanced_skill

Overview

This skill is an advanced AI text humanizer for blog content that detects and removes 34 common AI writing patterns and adds personality to make copy read as genuinely human. It includes crypto/Web3-specific tells, severity scoring, stat attribution fixes, and a batch mode for processing multiple files. Use it to clean drafts, articles, and marketing posts that feel formulaic or machine-generated.

How this skill works

The humanizer scans text for 34 defined patterns, categorizes each match as HIGH, MEDIUM, or LOW severity, then offers a report showing line-level examples and an estimated edit count. With user approval it rewrites problematic passages, fixes vague statistics and attributions, and injects personality and tone tailored for general or developer audiences. A final readability check and em dash regression scan ensure edits stay natural and within target grade levels.

When to use it

• You suspect a draft sounds formulaic, robotic, or overly polished
• Preparing blog posts, articles, or white papers for publication
• Cleaning crypto or Web3 content with industry-specific AI tells
• Batch-editing multiple markdown drafts before release
• Verifying and fixing vague or unreferenced statistics

Best practices

• Run a --scan first to review issues before committing fixes
• Accept HIGH severity fixes automatically; review MEDIUM/LOW items
• Provide source links for any stats the tool flags as unsubstantiated
• Choose target readability (grade 8–10 for general, 10–12 for developer)
• Use batch mode for bulk cleanup but spot-check outputs afterward

Example use cases

• Humanize a marketing blog post that uses hype words like 'revolutionizing' or 'seamless'
• Remove chatbot artifacts and knowledge-cutoff language from a technical draft
• Fix vague attributions such as 'research shows' by adding sources or rewording
• Process a folder of guest posts in batch to ensure consistent, human tone
• Clean Web3 content that makes unsupported decentralization or hype claims