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

Annotations already declare read-only, idempotent, non-destructive behavior. The description adds valuable behavioral context beyond that: it lists exactly what metrics are returned, explains that a `reference` causes scoring against ground truth, and that providing `model`+`api_key` triggers an LLM judge call. It also notes 'No re-execution' and the schema notes the API key is 'never stored,' which bolsters transparency.

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 description is compact—five sentences that front-load the core purpose, then layer conditional behavior and end with a clear constraint. Every sentence earns its place; no filler or redundancy.

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?

Given an output schema, rich annotations, and 100% schema parameter coverage, the description is complete enough. It covers the two major optional modes (reference-based scoring and LLM judging), the deterministic metrics, and the no-re-execution constraint, leaving no significant gaps for a tool of 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 coverage is 100% with good per-parameter descriptions, yielding a baseline of 3. The description adds semantic value by explaining how parameters interact: `reference` makes each output scored against it, `model`+`api_key` enables the judge, and the metric list (token cosine, ROUGE-L, Jaccard, JSON diff) clarifies what the tool computes from `response_a` and `response_b`.

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 opens with a specific verb and resource: 'Compare two ALREADY-PRODUCED outputs ... side by side.' It clearly differentiates from siblings by emphasizing no re-execution and listing concrete deterministic metrics plus an optional LLM judge verdict.

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 clearly states when to use the tool: when outputs already exist and you want a side-by-side comparison. It gives conditional guidance for `reference` and `model`+`api_key`, and explicitly says 'No re-execution — you bring the outputs,' which implies it is not for generating outputs. It does not name alternatives explicitly, so it's strong context without formal exclusions.

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