team-builder

Parallel multi-agent orchestration is executed by dynamically discovering persona markdown files, letting you select domains or specific agent names, and dispatching up to five agents simultaneously via the Agent tool.

The best way to select AI personas is through dynamic agent discovery, which probes project-local and global directories for markdown files, deduplicates by name, and allows flexible domain-based or fuzzy inputs like “1,3” or “security + seo”.

Multi-agent synthesis handles agreements and tensions by running selected agents in parallel, then generating unified reporting that summarizes findings by each agent and highlights agreements, conflicts, and next steps.

Yes, you need YAML frontmatter containing name and description fields in your markdown persona files to enable dynamic probing, deduplication, and proper agent selection during team synthesis.

Parallel agent dispatch is limited to a maximum of five agents selected from deduplicated persona markdown files, ensuring focused multi-agent orchestration and unified reporting without overwhelming the Agent tool.

Yes, you can use fuzzy inputs to pick agents from different domains for workflow automation, allowing flexible selections like specific agent names, domain lists, or index numbers to build your parallel team instantly.