blader/humanizer

What it solves

Humanizer addresses the issue of "AI-sounding" text—prose that feels artificial, formulaic, or overly polished—by rewriting it to read more naturally, like a human wrote it, without altering the original meaning or factual content.

How it works

The tool utilizes a set of 35 specific patterns derived from Wikipedia's "Signs of AI writing" to identify and remove artificial markers. It follows a multi-step process:

1. Initial Rewrite: It performs a first pass to rewrite the text without being constrained by the original structure.
2. Critique and Refinement: It checks the draft against the 35 patterns and the original claims to identify remaining artificial elements.
3. Final Polish: It rewrites the remaining problematic sections.

It can also match a user's specific voice if provided with a writing sample, following the user's unique rhythm, word choice, and punctuation. When processing files, it targets only the prose, leaving code, data, and metadata intact.

Who it’s for

It is designed for anyone using AI agents to generate text who wants the output to sound more authentic and less like a chatbot, including writers, technical authors, and individuals creating personal content.

Highlights

• Pattern-Based Approach: Uses 35 distinct content, language, style, and chatbot patterns to strip away AI clichés.
• Voice Matching: Ability to mimic a user's personal writing style based on a provided sample.
• Factual Integrity: Strictly avoids inventing facts; all names, dates, and details must come from the source.
• Agent Integration: Works as a Markdown-based skill that can be integrated into agents like Claude Code or Claude Desktop.