Multi-agent Systems

Multi-agent systems need clear ownership boundaries, coordination contracts, memory surfaces, and evaluation criteria.

Working Principle

Prefer small, inspectable agent roles over opaque automation that cannot explain its handoff.