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

Annotations already declare readOnly, idempotent, non-destructive, closed-world, so the safety profile is covered. The description adds real context beyond that: the decision is deterministic, results are cost estimates rather than billing records, and no prompt content is consumed. It omits how the eight tuning knobs (budget_gate_usd, cache_hit_rate, retry_rate, etc.) alter the decision, which keeps it below a 5.

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?

Two front-loaded sentences: the outcome space first, then the estimation caveat and the input-privacy constraint. No filler, and the most decision-relevant information leads.

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?

The return shape is described (decision plus estimated costs plus fallback model), which is necessary since no output schema exists. But for an 11-parameter tool with 18% schema coverage, an agent has no guidance on how the optional cost-model parameters interact or which defaults govern the decision, leaving a meaningful gap.

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 only 18% — just 'model' and 'request_id' are documented — so the description must compensate and largely does not. It mentions model name and token counts, but the eight optional behavioral parameters (budget_gate_usd, cache_hit_rate, routed_cache_hit_rate, cold_start_cache_write, retry_rate, pass_rate, allow_async_batch) are left entirely unexplained in both schema and description.

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?

Names a specific verb and resource ('check one LLM request's token counts against a pricing snapshot') and enumerates the exact decision space: allow, pre-cliff warning, hard-cliff reroute, budget-gate block. This is clearly distinguishable from sibling read tools like list_supported_models and get_audit_evidence without opening the schema.

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?

'Estimates only, not billing' establishes the scope boundary and 'Receives model name and token counts only, never prompt content' implies a pre-flight check, but the description never states explicitly when to call this versus list_supported_models or the search/audit siblings. Usage is inferable rather than declared.

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