Сводка по клиентам

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

Annotations already declare readOnly, idempotent, non-destructive, so the safety profile is free; the description instead adds non-obvious operational traits: archival campaigns excluded from counts, per-client errors isolated into that row rather than aborting the call, a ~1 report/10s rate limit (~5 min for 30 clients), and a statsSkipped marker for stragglers. It also discloses API-point cost per client and the 50-client balance batching. Only gap: it does not describe response ordering or how the row shape looks beyond the fields mentioned.

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?

It is one long paragraph, but front-loaded with the purpose and the core semantics before cost details. Almost every sentence carries a decision-relevant fact (fast vs slow path, error isolation, point cost). It could be split into purpose / cost / output sections for faster scanning, which keeps it from a 5.

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?

No output schema exists, so the description carries the return-shape burden and does: per-client general-account balance, campaign counts split by state (on/suspended/off/ended) and moderation status, plus per-row error text and statsSkipped. Combined with disclosed latency and cost, an agent has everything needed to choose and call it correctly.

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?

Schema coverage is 100%, so the 3 baseline applies, but the description genuinely adds semantics: clients come from list_clients unless an explicit logins list is given, stats pulls impressions/clicks/spend with VAT in the client's own currency, and the 5-minute ceiling explains why statsSkipped appears. The limit parameter's interaction with traversal is 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?

The first sentence names a specific resource and scope: an aggregate summary over all agency clients and managed accounts in one call. It explicitly differentiates from siblings by naming list_clients as the source of clients and warning against looping list_campaigns per client. An agent can distinguish this from get_balance, get_statistics, and list_campaigns without opening any 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?

It gives a concrete triggering question («что у меня по аккаунтам в управлении») and an explicit conditional: request stats only when the question is about money or traffic, otherwise the fast path applies. It also names the anti-pattern alternative (looping list_campaigns) and states this summary is cheaper, so the when-not is covered.

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