AgentDrop Inference

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

Goes far beyond the readOnlyHint annotation: it discloses the exact pricing formula and minimum charge, that the 402 carries the exact amount, that billing only occurs on successful model calls, that prompts are forwarded to the provider and not stored, and that there is no streaming. These are the operational facts an agent needs before committing to a paid call.

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 capability and default model, then moves to pricing and model-specific caveats, with no filler sentences. It is dense and reads as one long block rather than clearly separated concerns, which slightly hurts scanability, but every clause carries information.

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 10-parameter paid inference call with no output schema, the description covers billing, defaults, model routing, reasoning-token behavior, and data handling. The remaining gap is that it never sketches the return shape (choices/message/usage), which matters when no output schema exists, and four parameters remain undocumented.

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 0%, so the description must compensate, and it does for model (with default), messages (OpenAI-compatible shape), max_tokens (default 1024, max 8192), temperature, tools, and reasoning_effort (default low). However stop, top_p, tool_choice, and response_format are documented in neither place, leaving roughly half the parameters opaque.

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 and resource ('Chat completion on a frontier model') and immediately enumerates the available models plus the default, so the agent knows exactly what this does. It also names the sibling 'models' tool as the place to discover model options, distinguishing the two.

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?

Explicitly routes model discovery to the 'models' tool and gives conditional guidance for thinking models ('give them room, or set reasoning_effort'). It also states the message shape is 'Same shape as OpenAI chat,' which tells the agent how to construct calls. Lacks an explicit statement of when not to use this tool, but that case is thin.

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

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

readOnlyHint=true already establishes the safe-read profile, so the description only needs to add beyond it. It does: 'Free' discloses that the call consumes no credits, and 'limits' signals the response includes quota/context constraints. Minor gap: no statement of whether the list is cached or how large it is.

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?

One sentence plus a one-word cost flag, with the resource front-loaded and the payload contents enumerated. Nothing is wasted and nothing essential is buried.

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 no-parameter, no-output-schema, read-only catalog tool, the description covers what is returned (models, rates, formula, limits) and the cost profile. It needn't explain return structure, though a hint about whether pricing is per-token or per-request would close the last 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?

Zero parameters, so there is nothing for the description to disambiguate; baseline 4 applies. The description correctly implies no filtering input is required — the tool returns the full catalog.

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?

Specific verb + resource: 'List the available models', and it further names the payload (provider rates, pricing formula, limits). An agent immediately knows this is a catalog/metadata lookup rather than an inference call. It does not explicitly differentiate from the only sibling, chat, but the noun 'models' already does much of that work.

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?

Usage is implied — call this to discover models and pricing before/independently of chat — but never stated as a when-to-use rule and no alternative is named. 'Free' hints at a zero-cost lookup path, which is suggestive rather than directive.

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