team-builder

You can assemble agent teams by selecting up to five markdown-based agent personas to execute in parallel. The skill discovers available agents across flat or domain-subdirectory layouts, spawns them simultaneously, and synthesizes their isolated outputs into a unified report.

Parallel agent synthesis is the process of executing multiple markdown-based agent personas simultaneously and merging their results. The skill orchestrates discovery and parallel execution of agents with isolated prompts, then generates a final synthesized report from their combined outputs.

Agent discovery scans project directories for markdown personas organized in flat or domain-subdirectory layouts. The skill presents an interactive menu showing all discovered agents, allowing you to browse and select the optimal mix for your specific task.

Yes, you can select up to five agents and spawn them in parallel without manual coordination. The skill manages isolated prompts for each agent, executes them simultaneously, and automatically merges their individual results into a single synthesized report.

Markdown agent personas should include frontmatter controls to ensure safe parallel execution. The skill honors these frontmatter configurations and any optional resources specified, maintaining prompt isolation across all selected agents during simultaneous execution.

The best way to merge parallel agent results is through automated synthesis. After spawning selected agents with isolated prompts, the skill automatically collects their individual outputs and produces a unified synthesized report, eliminating the need for manual result aggregation.