What kind of feedback does the skill provide?

It provides a detailed report including an overall grade, specific strengths, issues categorized by severity, and a summary table of prioritized recommendations.

Do I need to provide a file path every time?

You can provide the path as an argument. If you don't, the skill will interactively ask you which file you would like to review.

How does the Prompt Reviewer fetch the latest best practices?

The skill uses the WebFetch tool to retrieve current documentation directly from Anthropic's official prompt engineering guide to ground its review in current recommendations.

Can I use this for modular instruction files?

Yes, it specifically evaluates the 'Composability' of instruction chunks to ensure they work effectively when injected into larger, dynamic prompt contexts.

Prompt Engineering Reviewer

Name: Prompt Engineering Reviewer
Author: Unsupervisedcom

byUnsupervisedcom

•

开发者工具

Reviews and optimizes prompt files against Anthropic's official engineering best practices to ensure high-quality AI outputs.

The Prompt Engineering Reviewer is a specialized Claude Code skill designed to audit and improve system prompts, instruction chunks, and agent definitions. By fetching the latest guidance directly from Anthropic’s documentation, it evaluates your prompt files against ten rigorous criteria—including structural integrity, role-based prompting, and handling of edge cases. This skill provides a structured feedback report with actionable recommendations, a quality grade (A-F), and specific line-by-line improvements, ensuring your AI instructions are clear, concise, and production-ready.

主要功能

01Generates structured reports with A-F grading and priority-based issue tracking.

02Analyzes the specific context of the prompt, whether it is standalone or an instruction chunk.

0335 GitHub stars

04Fetches real-time Anthropic prompt engineering documentation for up-to-date guidance.

05Provides specific before-and-after code suggestions for immediate prompt optimization.

06Evaluates prompts against 10 core criteria including clarity, structure, and examples.

使用场景

01Refining modular instruction chunks to ensure they are compatible with larger prompt contexts.

02Training and onboarding developers on prompt engineering standards through automated feedback.

03Auditing complex system prompts before deploying an AI agent to a production environment.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add unsupervisedcom/deepwork prompt-review

For use in Claude.ai and ChatGPT

Download Skill