Try the delegate first

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

Goes well beyond the annotations by disclosing cost ('it costs what those answers cost, $0 on free plans'), duration ('a few minutes'), the async contract (start returns at once, status carries progress/report), and a privacy guarantee ('Nothing of the user's is sent'), which is meaningful given openWorldHint=true. It also describes the report contents: lowest effort that passed plus the CLAUDE.md line.

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-loads the purpose, then the cost/duration, then the action semantics. Dense with semicolon-chained clauses but nearly every clause earns its place. Slightly overstuffed for a single-parameter tool.

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?

Covers the things an agent actually needs for an async, cost-incurring benchmark: what runs, how long, what it costs, how to poll, how to abort, and what the report contains; no output schema is present to carry that. Minor gaps remain (rate limits on repeated status polling, whether a stopped trial retains partial results).

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 already 100% and the enum is documented, so the baseline is 3. The description improves on that by explaining what each action does behaviorally — start fires and returns immediately, status yields progress or the finished report, stop terminates — plus confirming status is the default.

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 concrete verb and object: measure how well 'delegate' answers on the user's models, using 10 known-answer questions at three effort levels. That is far more specific than the name 'trial' and clarifies the sibling relationship by making 'delegate' the subject under test. It stops short of naming the alternatives (grade, estimate) explicitly.

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?

Gives a clear condition for use — 'before relying on it' — so the agent knows this is a pre-adoption check on delegate rather than a general run. It also maps the three actions to lifecycle stages (start/status/stop). No explicit when-not or named alternative is given, so it falls 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.