zircote/human-voice

6 stars · Last commit 2026-04-22

Detect and eliminate AI writing patterns in your content. This Claude Code plugin performs multi-tier analysis of character patterns, language cues, structural issues, and voice authenticity. Auto-fix em dashes, smart quotes, and emojis. Keep documentation and prose sounding genuinely human.

README preview

# Human Voice Plugin

[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT)
[![Claude Code Plugin](https://img.shields.io/badge/Claude_Code-Plugin-blueviolet)](https://github.com/anthropics/claude-code)
[![CI](https://github.com/zircote/human-voice/actions/workflows/ci.yml/badge.svg)](https://github.com/zircote/human-voice/actions/workflows/ci.yml)
[![Version](https://img.shields.io/badge/version-0.2.0-green.svg)](https://github.com/zircote/human-voice/releases)
[![Node.js](https://img.shields.io/badge/Node.js-18+-339933?logo=nodedotjs&logoColor=white)](https://nodejs.org)
[![GitHub Stars](https://img.shields.io/github/stars/zircote/human-voice?style=social)](https://github.com/zircote/human-voice)

A Claude Code plugin that detects AI-generated writing patterns and builds voice profiles through adaptive interviews and computational stylistics.

> **Experimental.** This project is a research prototype. The scoring pipeline, dimension mapping, and NLP analysis have not been validated against external benchmarks or peer-reviewed psychometric standards. The voice profiles it produces are plausible but unproven. Treat the output as a starting point for editorial guidance, not as a validated instrument. The question bank, scoring weights, and dimension definitions will change as the system matures. Use it, break it, report what does not work.

![Human Voice Plugin - 4-tier analysis flow](.github/infographic.jpg)

## Features

- **Multi-tier pattern detection**: Character, language, structural, and voice analysis
- **Automated character fixes**: Auto-fix em dashes, smart quotes, emojis
- **Proactive review**: Agent triggers after content creation/editing

View full repository on GitHub →