Kwai

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

Annotations already cover readOnly/idempotent/non-destructive/openWorld, and the description still adds substantive behavior: each call costs the credits listed per resource, the response reports what was charged and what remains, failures are free, and cached responses are free. It also discloses pagination semantics (list resources return a `cursor`, pass it back unchanged). This is exactly the kind of cost/side-effect context annotations cannot express.

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?

Structure is strong: a one-line purpose, an invocation-rule sentence, a cost/pagination sentence, then a scannable resource list with cost and required argument inline. Front-loading is good. The middle sentence compresses credits, cursor, and caching into one long semicolon chain, which slightly hurts readability but does not waste words.

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 7-parameter dispatcher with a nested free-form `params` object and no output schema, the description covers the essentials an agent needs to call it: resource choice, per-resource arguments, cost, caching, and pagination. The remaining gap is that `params` is a partial escape hatch whose contents are deferred to `describe`, so an agent cannot call less-common optional parameters without a second tool call. Given the tool's breadth, that is a minor rather than disqualifying omission.

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 baseline is 3; the description earns above that by encoding resource-to-parameter binding that the flat schema cannot express — profile and profile.posts accept `url|handle`, post accepts `url`, and profile.posts is the only paged resource. It also reiterates that `params` is for unknown-by-name optional arguments and points at `describe` for types and allowed values. It does not, however, enumerate the optional parameters it promises are 'in the tool description,' which is a small internal inconsistency.

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 description names the resource domain (Kwai profiles, posts, videos) and then enumerates each selectable resource with the argument each one needs, so an agent knows exactly what it can fetch. It is distinguishable from sibling platform tools (youtube, tiktok, instagram) by the explicit platform name. It stops short of a single crisp verb phrase — 'Data from Kwai' is a noun phrase rather than 'Fetch X' — but the resource list removes nearly all ambiguity.

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 clear operating instructions: pick a `resource` first, then pass that resource's parameters as top-level arguments, and put extras in `params`. It also gives cost-aware usage guidance — pass `max_age_days` to accept a cached copy for free — which is real decision support. The gap is sibling routing: it never says when to use this versus `get_profile`, `get_post`, or `search`, which is the most likely selection mistake for an agent.

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