Prompt Improver

Analyze and enhance vague or incomplete prompts before execution.

Usage

/prompt-improver [prompt]

Arguments:

• prompt - The original prompt to analyze and improve

Process

Step 1: Analyze the Prompt

Evaluate for:

• Clarity: Is the intent clear?
• Specificity: Are requirements concrete?
• Context: Is there enough background?
• Scope: Are boundaries defined?
• Success criteria: How do we know when it's done?

Step 2: Identify Gaps

Look for missing information:

Gap Type	Example
Ambiguous target	"Fix the bug" - which bug?
Missing context	"Update the API" - what change?
Unclear scope	"Improve performance" - where? how much?
No success criteria	"Make it better" - better how?
Missing constraints	"Add a feature" - any limitations?

Step 3: Research Context

Before asking questions, gather context:

• Check recent git changes
• Look at open files or selections
• Review related code
• Understand existing patterns

Step 4: Generate Clarifying Questions

Ask targeted questions to fill gaps:

To help you effectively, I need to clarify:

1. **Target**: [Specific question about what to change]
2. **Scope**: [Question about boundaries]
3. **Constraints**: [Question about limitations]
4. **Priority**: [Question about what matters most]

Keep questions:

• Specific and actionable
• Limited to 2-4 most important gaps
• Informed by context already gathered

Step 5: Reformulate the Prompt

Once clarified, rewrite as a clear, actionable prompt:

Original: "Fix the login"

Improved: "Fix the login issue where users with special characters in their password cannot authenticate. The bug is in the password validation logic in auth.ts. Success criteria: users can login with any valid password."

Output Format

## Prompt Analysis

**Original**: [original prompt]

**Clarity Score**: Low / Medium / High

**Gaps Identified**:
- [Gap 1]
- [Gap 2]

**Context Found**:
- [Relevant context from codebase]

**Clarifying Questions**:
1. [Question 1]
2. [Question 2]

---

After user responds:

## Improved Prompt

[Clear, specific, actionable version]

Guidelines

• Don't over-engineer simple requests
• Skip this for already-clear prompts
• Research before asking obvious questions
• Prioritize the most critical gaps
• Make the improved prompt self-contained