Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations cover non-readOnly, non-idempotent, open-world semantics, and the description adds real value on top: the returned auth_link must be opened by the user to complete authorization, and Pinterest connections require service_conditions_id and return no auth_link. It does not explain the two-step completion flow or what happens if the link is never opened.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.