Tool Design

This is prompt engineering for tool APIs, treating descriptions as direct inputs to agent reasoning rather than documentation for humans. The core insight is solid: if you can't definitively say which tool an agent should pick in a situation, the agent won't either. It pushes you to consolidate overlapping tools and write descriptions that answer what, when, and what returns, since those are exactly the questions agents evaluate during selection. The consolidation principle alone will save you from a lot of confusing multi-tool scenarios. Worth reading if you're debugging why agents keep picking the wrong tools or if you're designing APIs that LLMs will call directly.