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

Annotations already declare readOnlyHint=true, openWorldHint=false, destructiveHint=false, so the safety profile is covered. The description adds real value with a RETURNS line (AUC, optimal threshold, sensitivity/specificity), which matters because there is no output schema. It stays silent on edge cases such as tied scores, minimum sample size, or what happens with a single-class label vector.

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?

The pipe-delimited USE WHEN / ALIASES / RETURNS layout is front-loaded and scannable, with no filler sentences. The opening repeats the title verbatim ('ROC curve & AUC / ROC 曲线与 AUC'), which is wasted tokens given the title field already carries it.

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 two-parameter, annotation-covered, side-effect-free computation tool, the definition covers trigger condition, synonyms, and return values — essentially everything needed to call it. Only minor operational details (sample-size expectations, tie handling, output shape beyond field names) are absent, which is acceptable at this complexity.

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 100% (labels = true 0/1 labels, scores = predicted scores/probabilities), so the schema carries parameter meaning. The description only restates the same concepts in prose ('binary gold standard and predicted scores/probabilities') without adding format constraints such as equal-length arrays or score-range expectations. Baseline 3 applies.

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?

States a specific computation (ROC curve & AUC) plus the exact output concept set — discrimination and best cut-off — and the required data shape (binary gold standard plus predicted scores/probabilities). An agent can distinguish this from sibling stats tools such as classification_metrics or delong_auc_compare 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?

The 'USE WHEN' clause gives a clear triggering condition: binary gold standard plus predicted scores and a desire for AUC/ROC/discrimination/cut-off. The alias list further widens matching (cutoff, 诊断效能, 敏感度特异度). It does not name a when-not condition or point to alternatives like delong_auc_compare for comparing two AUCs, so it stops short of a 5.

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