A skill file designed to help Large Language Models (LLMs), particularly Claude, identify and remove common "AI tells" from prose. It aims to improve the authenticity and human-like quality of AI-generated text, benefiting writers and content creators seeking more natural-sounding output.

How It Works

The core of the project is a structured set of instructions (SKILL.md) and reference files (phrases.md, structures.md) that guide LLMs to avoid predictable patterns. It targets specific categories of AI writing artifacts: banned phrases (e.g., throat-clearing openers), structural clichés (e.g., binary contrasts), and stylistic habits (e.g., tripling). This approach leverages pattern recognition to refine AI-generated prose.

Quick Start & Requirements

Integration is straightforward for Claude users: add the skill folder directly, upload SKILL.md and reference files to project knowledge, or copy core rules into custom instructions. For API users, include SKILL.md in the system prompt. Reference files are loaded on demand. No specific hardware or non-default software prerequisites are mentioned beyond access to an LLM like Claude.

Highlighted Details

The skill identifies and helps eliminate specific AI writing patterns, including banned phrases, structural clichés, and stylistic habits. It includes a scoring mechanism, rating prose on Directness, Rhythm, Trust, Authenticity, and Density (out of 50). Scores below 35/50 indicate a need for revision, providing a quantitative measure for prose quality.

Maintenance & Community

No information regarding contributors, community channels, or project roadmap is available in the provided README.

Licensing & Compatibility

The project is released under the MIT License, permitting free use and wide sharing. This license generally allows for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

The skill is explicitly designed as a "skill file" for Claude, though it can be adapted for other LLMs. Its effectiveness may depend on the LLM's ability to interpret and apply the provided instructions and references. No specific limitations on platforms or features are detailed.