council

Multi-agent codebase exploration works by spawning a configurable number of task agents, defaulting to n=3, to investigate a specified area. These agents apply diverse perspectives to surface architecture, keywords, and hotspots, consolidating findings into a synthesized summary.

For an architecture review, the skill orchestrates multi-agent exploration of a feature area to surface architecture, keywords, and hotspots. It scales to the area's size and complexity with configurable agent counts, ensuring thorough yet controlled exploration.

Yes, an agent framework is required to run this multi-agent codebase exploration. The skill relies on an agent framework to spawn and coordinate the sub-agents needed for mapping architecture and hotspots, using deterministic defaults with safe fallbacks.

Yes, you can configure the number of agents for a codebase audit. The skill spawns three task agents by default, with optional variation to ensure diverse perspectives, and scales the agent count based on the area's size and complexity.

When scaling sub-agents for large codebase areas, the skill uses validation and safe fallbacks to ensure thorough yet controlled exploration. While agent count is configurable to match area complexity, it depends on the underlying agent framework's capacity.