Prompts

First Pass: Purpose and Activation

Clear Purpose Statement: Every prompt/command file should answer these questions in first 1-3 lines:

Examples:

✅ GOOD - Clear purpose:

❌ BAD - Vague purpose:

Second Pass: Completeness

Complete Prompt Elements:

• Not done: Clear description of task
• Not done: Expected input format (if applicable)
• Not done: Expected output format
• Not done: Examples (if task is complex)
• Not done: References to relevant skills or documentation

Self-Contained vs Skill Reference:

✅ GOOD - Simple task, self-contained:

✅ GOOD - Complex task, reference skill:

❌ BAD - Complex task without guidance:

Third Pass: Prompt Engineering Quality

Clarity and Specificity:

❌ VAGUE: “Look at the files and find problems” ✅ SPECIFIC: “Analyze modified Kotlin files for MVVM violations: mutable state exposure, improper dependency injection, missing error handling”

Logical Structure:

✅ GOOD - Step-by-step:

Examples for Complex Tasks:

✅ GOOD - Shows expected format:

Fourth Pass: Session Context

Context Requirements:

✅ GOOD - Explicit about requirements:

❌ BAD - Assumes context without stating:

Graceful Handling:

✅ GOOD - Handles missing info:

Fifth Pass: Skill References

Proper Skill References:

✅ GOOD - Clear skill invocation:

Verify:

• Not done: Referenced skill exists in .claude/skills/
• Not done: Skill name matches exactly (case-sensitive)
• Not done: Command adds value beyond just calling skill
• Not done: Command purpose is distinct from skill purpose

Priority Classification

Classify findings using reference/priority-framework.md:

• CRITICAL - Prevents functionality or exposes security vulnerabilities
• IMPORTANT - Significantly impacts quality or maintainability
• SUGGESTED - Improvements that aren’t essential
• OPTIONAL - Personal preferences