对比某 identity 的最近两次观测，判定提升/退化/中性（带阈值与样本门槛说明）。
判定门槛：相对变化 >= 20% **且** 绝对变化 >= 0.05 才算"显著"（severity: rise / sharp_drop）。
样本 n < 20 一律返回 severity=low_n，**不给出升降结论**——这是防误报的硬门。
Args:
identity: 品牌名或域名
kind: 观测类型，留空取全部
series: 题集/口径标识，留空取全部
| Name | Required | Description | Default |
|---|---|---|---|
| kind | No | ||
| series | No | ||
| identity | Yes |
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |
Changes observed during successful MCP inspections.
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, idempotentHint, non-destructive and open-world, so safety is covered. The description goes well beyond them by disclosing the significance rule (>=20% relative AND >=0.05 absolute) and the hard n<20 suppression to severity=low_n, which materially affects how an agent should interpret the output. Missing only edge behavior such as a single-observation case.
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: purpose sentence first, then the decision thresholds, then the arg list. Every section is short and functional. The Args block restates parameter names, but that restatement is necessary because the schema has no per-property descriptions.
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 read-only diff tool with an output schema present, return structure need not be described, and the description instead supplies the evaluative semantics (severity values, thresholds, low_n). Remaining gap is the degenerate case of fewer than two observations and whether comparisons span the requested kind/series filter or span across them.
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 carries the full burden and does document all three parameters: identity as brand name or domain, and kind/series as optional filters that default to 'all' when blank. It does not enumerate legal values for kind or give a format example for series, which keeps it short of a 5.
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 verb (diff/compare) plus resource (observations) and narrows scope precisely to 'the most recent two observations of an identity'. An agent can distinguish this from siblings like query_history, which returns history rather than a two-point comparison, without opening either 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 description implies usage via the two-observation comparison scope and the low_n gate, which tells the agent what results to trust. However, it never states when to prefer this over query_history or score_visibility, nor any preconditions (e.g. needing at least two observations to exist). Implied context only.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.