Principlev1
Decompose complex recurring behaviors into narrow-scope
Decompose complex recurring behaviors into narrow-scope agents with clear interfaces rather than building multi-purpose agents, as complexity should live in agent interactions not in individual agent logic.
Why This Is a Principle
Derives from hierarchical chunking (Hierarchical Chunking Expands Capacity) and society of mind (The mind operates as multiple semi-autonomous processes). The principle prescribes architectural constraint: narrow scope per agent. It's actionable (tells you HOW to decompose), general (applies to any agent system design), and maps to both Minsky's architecture and modern multi-agent AI systems discussed in the lesson.