What problem does it solve? Getting genuinely diverse perspectives from multiple AI agents is hard when a single LLM roleplays every character, because the opinions tend to converge and feel performative. This Skill solves that by spawning each BMAD agent as a real, independent subagent so every voice thinks for itself. ## Core Features & Use Cases - Independent Subagent Spawning: Each selected agent runs as its own subagent process via the Agent tool, producing authentic disagreement and domain-specific insight. - Dynamic Roster Resolution: Loads the available agent roster from layered TOML config files and project context, then picks 2-4 relevant voices per round. - Flexible Modes: Supports a --model flag to control subagent model weight and a --solo flag to roleplay all agents in a single response when subagents are unavailable. - Use Case: You are debating a technical architecture decision for your project. Invoke party mode to have the architect, developer, and product manager agents each give their unfiltered take, then ask one agent to respond directly to another's critique. ## Quick Start Ask the AI to start party mode so the installed BMAD agents can discuss your project topic from multiple perspectives.
