Does this work with languages other than Python?

Yes, the skill assesses the logic, documentation, and requirement fulfillment of any text-based output regardless of the programming language used.

What happens if the output fails the assessment?

The tool provides a 'verdict' such as ITERATE or RESTART, along with a list of specific gaps that need to be addressed in the next iteration.

Can I change the passing grade for my code?

Yes, you can use the --threshold parameter to set your own quality bar, such as 0.95 for critical systems or 0.85 for general utilities.

What does the quality score represent?

It is a normalized value between 0.0 and 1.0 that represents how well the output satisfies your task requirements and general production standards.

How can I use this to automate my workflow?

You can chain it with other commands to create a loop that checks the output score and automatically triggers a re-prompt if the score falls below your threshold.

Output Quality Assessor

Name: Output Quality Assessor
Author: HermeticOrmus

byHermeticOrmus

•

Seguridad y Pruebas

Evaluates LLM-generated code and content against specific requirements to provide a structured quality score and actionable feedback.

The Assess Quality skill provides a standardized framework for measuring the excellence of AI-generated outputs within the Claude Code environment. By scoring work across four critical dimensions—correctness, completeness, clarity, and production quality—it eliminates guesswork in the development lifecycle. This tool is essential for developers building automated iteration loops or CI/CD pipelines where a specific 'quality bar' must be met before code is accepted, offering detailed insights into strengths, gaps, and specific recommendations for further improvement.

Características Principales

01Automated verdicts to streamline 'Accept' vs 'Iterate' decision-making

02Multi-dimensional scoring across correctness, completeness, clarity, and quality

03Structured XML feedback including strengths, gaps, and iteration recommendations

04Seamless integration with terminal-based workflows and meta-prompting chains

054 GitHub stars

06Customizable quality thresholds to enforce project-specific standards

Casos de Uso

01Benchmarking different LLM prompts to identify which produces the most reliable logic

02Building self-correcting AI loops that automatically iterate until a quality threshold is reached

03Automating code reviews to ensure AI-generated snippets meet production standards

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hermeticormus/claude-code-game-development assess-quality

For use in Claude.ai and ChatGPT

Download Skill