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

With no annotations provided, the description carries the full burden and does so well: it declares statelessness (no memory, independent requests), that no free text is produced, the answer payload shape (selected option, confidence, probability over all options), and billing semantics (input-token based, ~2 CNY per million, output free, failures not billed). These are exactly the operational facts an agent cannot infer from the schema.

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?

Content is front-loaded: purpose first, then applicability, then parameter format, then example, then billing. Given 0% schema coverage, the long criteria-format sentences and the JSON example earn their space. Minor cost is that the type-format enumeration is dense prose that a bulleted list would scan faster.

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?

For a complex, nested, stateless inference tool with no annotations, the description covers purpose, use cases, the full question-type grammar, the answer shape, and cost. An output schema exists, so the probability payload need not be detailed further. The remaining gap is the `state` input contract and any size/validation limits on the 1-32 question range.

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 description coverage is 0%, so the description must compensate — and it largely does for `questions`: the 1-32 cardinality, the {question_id: question} mapping shape, the three types (choice/noul/score) with the exact `criteria` format for each, and an optional `instructions` field, all reinforced by a worked JSON example. The `state` parameter, however, is only glossed as '业务状态' with no guidance on what structure or size it accepts, leaving one of two parameters under-specified.

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 opening line states a precise contract: given a business state plus typed questions, it returns per-question calibrated probabilities and explicitly does NOT generate text. It even names its category ('System One 类型化判断'), which distinguishes it at a glance from every sibling (all TTS/ASR/OCR/document-task tools) without needing schema inspection.

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 names concrete application scenarios (ticket classification, risk decisions, QC evaluation) and a general condition ('structured judgment scenarios'), plus it warns there is no context memory so each call must be self-contained. It does not state when NOT to use it, but no sibling operates in this decision domain, so the absence of an alternative route is not a real gap.

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