team-builder

Multi-agent orchestration coordinates multiple agent personas into on-demand teams to drive parallel problem solving. It dynamically selects agents based on inferred domains, spawns them simultaneously via the Agent tool, and synthesizes their individual outputs into a final report.

Yes, the team builder supports both flat and domain-subdirectory agent collections. It uses dynamic domain inference to automatically categorize and select agents, allowing you to compose specialized teams across different organizational layouts seamlessly.

The best way to validate a web app across domains is to assemble a specialized team, such as Security and SEO agent personas, and run them in parallel. This simultaneously executes tasks and consolidates results for faster decision making.

Agent orchestration handles errors during parallel execution through per-agent error handling. This isolates failures so that if one spawned agent encounters an issue, it is managed independently without halting the broader workflow or preventing the final result synthesis.

No, manual domain specification is not required when building a multi-agent team. The system features dynamic domain inference that automatically detects the relevant domain context from your task to select appropriate agent personas from your available collections.