meta-prompt-creator

Design a system prompt for multi-agent teams by defining each agent's role, goals, tools, and constraints to ensure reliable and composable interactions. The Skill uses an interview-driven approach to gather this context and generates production-ready prompts tailored for orchestrator or subagent architectures.

The best way to structure a meta-prompt for an AI orchestrator is using an MCWOH-based template that enforces YAML frontmatter with name and description. This provides architecture guidance and includes safety constraints to keep the orchestrator focused on routing tasks to subagents effectively.

Yes, you can use this prompt engineering approach for solo agents by providing the agent's role, goals, tools, inputs, and constraints. It generates tailored single-agent system prompts using MCWOH templates, ensuring the agent remains focused and production-ready without needing a team structure.

Yes, you need to provide team context to generate an AI agent system prompt for multi-agent workflows, as it determines the architecture pattern. If you are building a solo agent, team context is not required, and the prompt design focuses strictly on individual role and constraints.

Recommended architecture patterns for multi-agent prompt engineering include solo, orchestrator, subagent, dual-role, and pipeline patterns. The specific pattern is selected based on your stated task requirements and team structure to ensure proper workflow composition and reliable agent behavior.