Optimize performance

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

Annotations already indicate this is a mutating tool (readOnlyHint=false, destructiveHint=true), and the description substantiates that by noting that with apply=true it 'mutates flagged TOP resolutions by scale and returns the before/after sizes.' It also clarifies the default safety posture (read-only unless apply=true). However, it doesn't detail specific destructive side effects (e.g., whether mutations can be reverted, whether it permanently overwrites original TOPs) beyond the annotations, and it doesn't mention rate limits or authorization requirements. Given the annotations already carry the core safety signal, the description adds meaningful context without fully disclosing all mutation consequences.

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 three sentences with no fluff. The first sentence front-loads the core purpose, the second qualifies the mutation behavior and return info, and the third provides explicit sibling guidance. Every clause carries information—there are no wasted words or filler phrases like 'this tool is designed to' or 'it should be noted that.' It is the ideal size for an agent to parse quickly.

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 tool with 4 self-documenting params, 100% schema coverage, and a clear output described as 'returns the before/after sizes,' the description covers the essentials: operation, mutation flag, safe default, and sibling differentiation. There is no output schema, but the description states the before/after return values, which compensates. The only gaps are details about error conditions, what happens if there are no flagged nodes, or whether the returned sizes are filesystem sizes vs. resolution sizes, but these are minor for an agent selecting and invoking the tool.

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?

The input schema already covers 100% of parameters with descriptions, so the baseline is 3. However, the description adds valuable semantic context: it explains that apply=true 'mutates flagged TOP resolutions by scale' and that the default false is a 'plan-only pass,' effectively tying the boolean and scale parameters together. It also gives concrete examples for scale ('0.5 = half on each axis'), which enriches the schema's bare number constraint. The description does not restate schema fields verbatim; it amplifies their intent.

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 'Scan a network for cook-time bottlenecks and report the slowest nodes with concrete suggestions,' which captures exactly what the tool does. It clearly distinguishes this from the sibling 'get_td_performance' by explaining that this tool is for bounded resolution changes while the sibling is for metrics-only. The verb 'scan and report' plus the mutation qualification with apply=true makes the scope unmistakable.

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 explicitly states when to use this tool vs. the alternative: 'Run get_td_performance when you only need metrics; use this tool when you want the bounded resolution change, and leave apply=false for a plan-only pass.' This is textbook guidance—it names the sibling, gives a concrete decision rule, and clarifies the safe default (apply=false) for a plan-only pass.

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