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

Annotations already declare readOnlyHint, idempotentHint, and destructiveHint=false; the description adds genuine context beyond them by stating no code runs, no external agent calls, no changes or deployment occur, and that synthetic prices are unsigned estimates never approved savings. That clarifies output trustworthiness, a real behavioral trait. Minor gap: no mention of latency or payload-size limits.

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 with the core action, then mode guidance, then the important caveat about description-only inputs, then safety disclaimers. Sentences are dense and each adds information; the trailing demo URL is a small non-essential addition but harmless.

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?

With no output schema, a nested open payload, and vague sibling boundaries, the description covers safety and a few payload modes but not the return shape or required inputs for a real evaluation. An agent could construct the simple sample_mode call but would be guessing for labeled baseline/candidate inputs.

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 0% and the single payload parameter is an open object with additionalProperties true, so the description must carry the semantics. It does name three usable payload modes, which is real added value, but leaves the shape of keys, required fields, and the labeled baseline/candidate format undefined.

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 states a specific verb and resource: run the Forge evaluator and optimizer on synthetic examples or supplied baseline/candidate outputs. It distinguishes itself from siblings like run_check or explain_verdict implicitly by naming evaluation plus cost optimization. However, it never explicitly contrasts itself with the closest sibling, so an agent must infer the boundary.

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 concrete modes (sample_mode: improved, regression, changed_path) and states that a description alone yields an integration plan rather than a measured evaluation, which is a useful when-not condition. But it never names alternatives such as run_check or list_scenarios, so the choice among siblings is only implied.

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