bmad-party-mode

Multi-agent roundtable discussions are orchestrated by spawning 2–4 independent subagents in parallel. Each agent produces an authentic, independent perspective on the same topic, which is then presented to the user for cross-domain project reviews.

Spawning independent subagents ensures each agent produces an authentic, independent perspective in parallel. Roleplaying multiple voices in a single model response is a solo fallback mode used when subagents are unavailable or when speed matters more than independence.

Yes, you can optionally incorporate project context files when orchestrating agent discussions. The Skill parses invocation arguments to load a configured agent roster and integrates context to coordinate parallel Agent tool calls with bounded discussion summaries.

Yes, Python3 is required as a dependency to facilitate multi-agent discussions. The environment must support spawning real subagents and coordinating parallel Agent tool calls to manage the roundtable orchestration process.

Follow-up reactions in a multi-agent brainstorming session are managed by keeping a bounded discussion context summary. This supports targeted reactions, allowing you to ask specific agents what they think about statements made by others.

Solo roleplay mode should be used when subagents are unavailable or when execution speed matters more than independent perspectives. It roleplays all selected agents within a single model response rather than spawning parallel subagents.