Refine

---
description: Improve and enhance existing agents, commands, or prompts using expert prompt engineering principles
---

# Refine Agents, Commands, or Prompts

Use the prompt-engineering agent to transform basic prompts into comprehensive, context-rich instructions or improve existing agents and commands.

## Task

I need help refining: $*

## Context

The refinement process will:
1. **Analyze** the current prompt/agent/command structure
2. **Identify** gaps, ambiguities, and improvement opportunities
3. **Enhance** with proper structure, examples, and actionability
4. **Validate** against quality standards and best practices

## Requirements

Use the @task prompt-engineering agent to:

### For Agents
- Review role definition and core mission
- Enhance methodology and process steps
- Add concrete examples and scenarios
- Improve quality standards and principles
- Optimize tool selection and model choice
- Ensure clear invocation triggers

### For Commands
- Review namespace organization
- Enhance argument handling
- Add usage examples
- Improve agent delegation (if applicable)
- Ensure documentation completeness
- Validate consistency with conventions

### For General Prompts
- Transform into structured, actionable format
- Add role definition and responsibilities
- Include methodology and process steps
- Provide concrete examples
- Define quality standards
- Maintain professional tone

## Output Expectations

The agent will provide:
- ✅ Comprehensive analysis of current state
- ✅ Specific improvement recommendations
- ✅ Enhanced version with full structure
- ✅ Explanation of changes and rationale
- ✅ Validation against quality checklist

## Examples

### Refining an Agent
```
/meta:refine The postgres-optimizer agent needs better database indexing strategy guidance
```

### Refining a Command
```
/meta:refine The /docs:update command should check for broken links and outdated screenshots
```

### Improving a Basic Prompt
```
/meta:refine "Act as a technical writer for API documentation"
```

## Quality Standards

All refined content will:
- Follow established prompt engineering patterns
- Include concrete, actionable instructions
- Provide real-world examples and scenarios
- Maintain professional, authoritative tone
- Optimize for context efficiency
- Align with Claude Code conventions