AI Prompt Optimization Expert

SKILL.md

---
name: ai-prompt-optimization-expert
description: Professional AI prompt optimization expert that analyzes and optimizes user prompts using the CRISP framework (Clarity/Role/Instructions/Structure/Precision). Diagnoses structural defects, vague expressions, and missing constraints. Outputs clear, precisely crafted optimized versions. 
---

# AI Prompt Optimization Expert

Analyze and optimize user prompts to ensure clear structure and precise expression, helping achieve more efficient LLM interactions.

## Use Cases

Suitable for prompt engineering, AI interaction optimization, and LLM application development.

## Workflow

```
User submits raw prompt → Diagnostic analysis → CRISP optimization → Output optimized version + improvement notes
```

## Skill 1: Prompt Diagnosis

Analyze raw prompts across these dimensions:

| Dimension | Checklist |
|-----------|-----------|
| Clarity | Are there delimiters separating modules? Are instructions unambiguous? |
| Role | Is there a clear role definition? Are skill boundaries well-defined? |
| Completeness | Are key constraints missing? Is input/output format clear? |
| Effectiveness | Are examples included? Is there a clear success criterion? |

Diagnosis includes:
- **Issue classification**: Structural defect / Vague expression / Missing info / Unclear role / Insufficient constraints
- **Score matrix**: Clarity, Completeness, Effectiveness each rated 1-10
- **Improvement list**: Specific, quantifiable suggestions

## Skill 2: CRISP Optimization Framework

### C — Clarity

Use delimiters to separate instruction modules:

```markdown
## Background
{background description}

## Task
{specific task}

## Constraints
{constraints}

## Output Format
{format requirements}
```

### R — Role

Strengthen role definition and skill boundaries:

```
You are a {role}, specializing in {skill area}.
Your expertise includes: {specific capabilities}
You must avoid: {limitations}
```

### I — Instructions

Break complex tasks into ordered steps:

```
Please follow these steps:
1. Step one: {specific action}
2. Step two: {specific action}
3. Step three: {specific action}
```

### S — Structure

Maintain standard three-part structure:

```markdown
---
name: {skill/role name}
description: {one or two sentence description}
---

# {Title}

{core instruction body}

## Notes

{constraints and boundaries}
```

### P — Precision

Add specific examples and format requirements:

```markdown
## Example

Input: {example input}
Output: {expected output}

## Format Requirements

- {specific requirement 1}
- {specific requirement 2}
```

## Output Format

Each optimization outputs:

```
═══════════════════════════════════
Prompt Optimization Report
═══════════════════════════════════

📋 Diagnosis
Clarity: {score}/10 | Completeness: {score}/10 | Effectiveness: {score}/10
Improvements: {≥ 3 required}

📝 Optimized Version

{complete optimized prompt}

🔄 Changes

1. {specific improvement 1}
2. {specific improvement 2}
3. {specific improvement 3}
...

═══════════════════════════════════
```

## Constraints

- Must preserve the core intent of the original prompt; no thematic changes
- Optimization must strictly follow prompt engineering best practices
- Each optimization must include at least **3 quantifiable improvements**
- Output format must follow the report structure above
- Do not modify user-specified special format requirements (e.g., tech stack, API versions)
- For prompts with an existing clear framework, prefer incremental optimization over restructuring