Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond the annotations (readOnly/idempotent/non-destructive already declare safety), the description adds real behavioral detail: nothing is stored or added to the user's book, amounts and account numbers are refused, and the map "restates the answers without judging them." That last point is genuinely useful tone/behavior guidance an agent cannot get from structured fields. It does not cover size limits or what happens at the 40-item cap, which is why it falls short of 5.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the output artifact, then the inputs, then the guarantees — the right order for an agent skimming. All content is relevant, but it is delivered as one long syntactic sentence, and the clause about institutions and countries a family would have to deal with is denser than it needs to be.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

An output schema exists, so return values need not be explained, and the description covers purpose, privacy posture, non-persistence and the meaning of the three optional answers. What remains unstated is edge behavior for large or incomplete place lists (the schema caps at 40 with a minimum of 1) and how this differs operationally from the data-driven geo tools.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Although top-level schema coverage reads 0%, the description supplies the practical meaning of each optional flag (owns or co-owns a business; sold a business/property or received a large sum in the last two years; whether one person holds the full list today) and constrains the places payload to names, kinds and countries only. This adds meaningful intent, especially the negative constraint against amounts and account numbers. The nested MapPlace fields themselves are left to the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

"Draws a one-page map of where the user's wealth sits, from what the user says" gives a specific verb (draws), a concrete artifact (one-page map), and a source of truth (what the user says) that separates it from data-backed siblings like get_geo_map or get_net_worth. The follow-on clauses spell out exactly what the map contains (grouped by country and kind, institution/country counts, what each institution sees). An agent can tell what this produces without opening the schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

"from what the user says" plus "Takes names, kinds and countries only, never amounts or account numbers" establishes a clear usage context: conversational intake of places, not a query against stored positions. The optional answers clause also signals when to pass owns_business, recent_liquidity and who_knows. However, no sibling is named and there is no explicit exclusion (e.g. use get_net_worth instead when real position data already exists), so routing still requires inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.