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

Adds substantial context beyond the annotations: partial-update semantics ('Only the fields sent are touched'), a validation gate (money facts refused without an approved training_spend record, spend bound to the approved cap), monotonic constraints on status and cleanup, and receipts merge behavior. The annotations only say non-read-only/non-destructive; the description carries the real behavioral profile.

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?

Every clause carries a rule rather than filler, and the fact list is front-loaded. The dense parenthetical asides and the trailing SCOPE sentence make it slightly harder to scan than ideal, but the size is proportionate to the number of rules.

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 9-parameter, nested-object mutation with no output schema and no annotations covering mutation semantics, the description covers gating, monotonicity, partial updates, and auth scope well. It omits what error/response behavior looks like on a refused or partial write, which is the one remaining unknown for an agent.

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 56%, and the description compensates by explaining the semantics of the gated fields — spend is capped to the approved record and status/cleanup only advance — which the schema enum alone does not convey. It does not cover customerId's superadmin-only targeting or the nested spend sub-fields, leaving a modest gap.

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 ('Report a job's independent facts') plus the exact resource and enumerates the fact families it writes (status, endpoint cleanup, spend, serving model id, provider job ref, receipts). This cleanly separates it from read-side siblings like get_training_job and creation-side like record_training_job.

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 explicit when-not guidance via scope: 'For the training runner's key. SCOPE: runner — held literally; write, admin and superadmin keys cannot call it,' and warns that 'The gate outcome is not settable here,' implicitly routing that to the gate-result tool. It never explicitly names an alternative tool to use instead for a given case, so it stops 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.