meta-prompt-generator_skill

Overview

This skill automatically generates structured, custom slash-commands (meta-prompts) from a short high-level description. It produces step-based workflows optimized for parallel execution, with dependency management, sub-agent strategies, and a comprehensive test suite. Outputs include execution modes, verification commands, and a ready-to-use command file.

How this skill works

Given a user prompt, the skill collects context (including web search for post-knowledge-cutoff topics) using parallel sub-agents, clarifies requirements through targeted questions, and designs a phased workflow that marks which tasks can run in parallel. It then generates the full slash-command: variables, directives, workflow steps, parallelization strategy, and automated test suites with validation commands.

When to use it

• You need a reusable, structured custom prompt or slash-command for a complex project.
• You want a workflow that maximizes parallel execution and clearly defines dependencies.
• You need autogenerated unit/integration/e2e tests and verification commands included in the prompt.
• You want to delegate sub-tasks to parallel sub-agents and coordinate their integration.
• You need a starting point for reproducible, version-controlled prompt-driven automation.

Best practices

• Provide a clear, specific high-level description to improve generated workflows.
• Iteratively test and refine the generated prompt before full automation.
• Version-control generated commands and track changes for reuse.
• Use the Task tool for independent tasks; keep integration in the main context.
• Include framework-specific verification steps (e.g., flutter analyze, pnpm build).

Example use cases

• Create a React+TypeScript TODO app prompt that parallelizes UI, API, and tests and saves the command file.
• Design a data pipeline prompt that splits load, clean, analyze, and visualization into parallel workers.
• Generate a prompt for a mobile feature that runs linting, unit tests, and build checks with a final sequential QA step.
• Produce a reusable slash-command that includes unit, integration, and e2e test suites for CI validation.
• Prepare a multi-agent research prompt that collects current web data via three parallel sub-agents and integrates sources.