How do you choose between exposing one powerful tool and several narrow tools?

Example Answer

I usually prefer several narrow tools when I care about reliability, permissions, and debuggability. Narrow tools make it easier for the model to choose correctly, easier for the host to authorize safely, and easier for engineers to understand what went wrong.

One powerful tool can look attractive because it reduces interface count, but it often pushes too much routing, validation, and business logic into one ambiguous action surface. That makes misuse more likely, not less.

I choose a broader tool only when the underlying operation is genuinely cohesive and the model can call it safely without hidden branching. The right question is not how many tools there are. It is how clearly each tool maps to one meaningful capability.

Common Poor Answer

A weak answer is saying one big tool is simpler because there is less for the model to learn. In practice, overpowered tools often increase ambiguity and risk.