kylehughes/writing-prose-like-a-human-for-agents

README preview

# Writing Prose Like a Human for Agents

[![Agent Skill](https://img.shields.io/badge/Agent_Skill-555?logo=claude)](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview)
[![Claude Code Plugin](https://img.shields.io/badge/Claude_Code_Plugin-555?logo=claude)](https://code.claude.com/docs/en/plugins)
[![Latest Release](https://img.shields.io/github/v/release/kylehughes/writing-prose-like-a-human-for-agents)](https://github.com/kylehughes/writing-prose-like-a-human-for-agents/releases)

*A skill and subagent that help AI agents write and edit prose that doesn't read like AI.*

## About

A Claude Code plugin with an Agent Skill and a Subagent for writing prose that avoids statistical patterns common to LLM output.

- **Agent Skill**: Vocabulary watchlists, structural checks, and before/after examples for identifying AI writing patterns.
- **Subagent**: Edits files in place to remove AI writing patterns. Uses the Agent Skill.

## Usage

The Subagent edits prose in place, applying five rules: cut significance inflation, use plain verbs, end sentences at the fact, vary rhythm, earn every adjective. Delegate to it from your main agent to keep prose edits out of your context window.

The Agent Skill has the vocabulary watchlist and rules. Use it directly when you want more control than the Subagent gives you.

View full repository on GitHub →