team-builder

To assemble an ad-hoc team of expert agents, you describe your task and interactively select up to five personas from different domains to spawn in parallel. The system then merges their synthesized results into a unified report for decision making.

Parallel agent execution asynchronously spawns multiple chosen agent personas simultaneously to process prompts. It synthesizes their independent outputs by merging the asynchronous execution results into a single unified report for downstream decision making.

Yes, you can use agent personas stored in domain-based subdirectory layouts. The system supports dynamic agent discovery across both flat file structures and domain-subdirectory collections, automatically inferring domains for interactive selection.

You source prompts for individual agents during team orchestration through per-agent prompt sourcing. The workflow supports providing specific prompt inputs for each selected persona before asynchronously spawning them for parallel execution.

The maximum number of agents you can spawn in parallel for a single task is five. You interactively choose up to five expert personas from various domains to form your ad-hoc team before asynchronous execution and result synthesis.