Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Adds a lot beyond the annotations, which only cover the read-only/idempotent safety profile: it discloses the extra values returned (section 382 rate, low-income housing credit percentages), the edge-case response for unissued months, and the returned ruling number and data_as_of. This is genuinely useful behavioral context for calling the tool correctly.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.