deployment_predict

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

Annotations already declare readOnlyHint=true, destructiveHint=false, and openWorldHint=true, so the description adds significant extra context: it discloses that returns include undocumented metadata fields, that costs follow the deployment's resource configuration (no per-request cost), and that cold starts may cause slow responses or 503 errors with advice to check health. These details go beyond the annotations and are highly relevant for an agent invoking the tool.

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 filler. It front-loads the core purpose, then covers return behavior, cost, and cold-start handling in a logical order. Every sentence contributes necessary information, making it highly efficient.

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?

Given the tool's complexity (no output schema, annotations cover safety), the description is complete: it explains what the tool returns (images/metadata verbatim, including undocumented fields), cost implications, and a specific failure scenario (cold start) with recommended action. It also covers all required parameters' semantics through the schema and the added default-owner note. Nothing critical for an agent to call it correctly is missing.

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 100%, so the baseline is 3. The description adds semantic value beyond the schema for the 'deployment' parameter by noting that 'owner defaults to the account owner' when a bare slug is used—information not present in the schema. It also reiterates that imagePath is local, matching the schema. For iou, conf, and imgsz, it adds nothing beyond the schema, but the additional default-owner behavior justifies a slightly higher score.

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 clearly states the action: 'Run inference through a deployment's own serving endpoint on a local image file'. It specifies the resource (deployment's endpoint) and the input (local image file), and clarifies how to reference the deployment (owner/deployment or bare slug with default owner). This distinguishes it from siblings like model_predict (model-based) and deployment_health (health check).

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 gives clear usage context: it explains how to specify the deployment and that the owner defaults to the account owner. It also provides a specific alternative: 'check deployment_health rather than retrying blindly' in the case of a cold start. While it doesn't explicitly contrast with model_predict, the deployment-specific nature is implied. The guidance is useful but could be more explicit about when to choose this over other inference tools.

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