Meta Prompt Generator

> Source: [bear2u/my-skills](https://github.com/bear2u/my-skills) | Category: C-AI-Engineering

---
name: meta-prompt-generator
description: Generate structured custom slash commands with parallel processing, dependency management, and test suites
---

# Meta Prompt Generator

## Overview

This skill takes a brief description and automatically generates structured custom slash commands (prompts) optimized for step-based parallel processing. It produces prompts with phased workflows, dependency management, parallel execution strategies, and comprehensive test suites. Use it when you need to turn a high-level request into a systematic, reusable Claude Code command.

## When to Use

- User requests a structured workflow for a complex project
- User needs a reusable custom prompt or slash command
- Task planning requires parallel processing optimization
- User requests prompts that include systematic testing

## Core Capabilities

### 1. Intelligent Knowledge Gathering
- Automatic web search for topics beyond the model's knowledge cutoff
- 3 parallel sub-agents per topic for knowledge gathering
- Integrated citations and verified information

### 2. Step-Based Workflow Design
- Tasks structured into logical phases
- Explicit dependency definitions between phases
- Independent tasks run in parallel; dependent tasks run sequentially
- Uses the `Task` tool for parallel processing strategy

### 3. Comprehensive Test Generation
- Auto-generates unit, integration, and end-to-end tests
- Identifies parallel execution feasibility per test
- Includes edge cases and validation commands

### 4. Optimized Execution Strategy
- Automatic identification of parallelizable tasks
- Sub-agent creation strategy
- Explicit execution modes (sequential / parallel / mixed)

## Workflow

After the user provides a high-level request:

1. **Context Gathering** (parallel)
   - Web search with 3 parallel sub-agents for unfamiliar topics
   - Parallel collection of relevant documentation

2. **Requirements Clarification** (sequential, user interaction)
   - Ask questions one by one when information is insufficient
   - Loop until all necessary information is collected

3. **Prompt Structure Design** (sequential)
   - Analyze parallelizable work areas
   - Determine inter-phase dependencies
   - Plan sub-agent strategy

4. **Content Generation** (sequential)
   - Write all sections in the specified format
   - Integrate searched knowledge
   - Structural validation

5. **Save and Report** (sequential)
   - Save to `.claude/commands/<prompt-name>.md`
   - Report generated prompt summary

## Generated Prompt Structure

The generated custom slash command contains these sections: