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

Adds substantial behavior beyond the annotations: credits are reserved and refunded in full on failure or block, max_credits prevents a job starting at a higher price than quoted, and the idempotency key prevents double-start/double-charge on retry. These are real operational traits (billing, retry safety, price drift) not derivable from readOnlyHint/destructiveHint/idempotentHint flags. Nothing contradicts the annotations — the key-based retry safety is consistent with idempotentHint=false meaning the tool is not naturally idempotent without the supplied key.

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 most consequential fact (credit reservation and refund policy) in the first sentence, then sequences prerequisites, mode choice, result delivery, idempotency, and price guard. Dense but with zero filler — every clause communicates a distinct constraint the caller must respect.

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, credit-consuming, nested-params mutation with no output schema, the description covers the failure modes that matter most: cost, prerequisites, retry safety, and result retrieval. It leaves based_on, consent_id, and webhook_url to the schema, which documents them adequately at 89% coverage. Slightly incomplete in not summarizing what a successful call returns (a job id to feed to get_job), which the caller must infer.

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 89%, so the schema already carries most parameter documentation; the description still adds genuine meaning for the highest-risk parameters. It explains prompt_mode's 'auto' vs 'exact' semantics and the maxLength constraint, the idempotency_key reuse rule, max_credits as a price guard, and wait_seconds as an alternative to polling. It does not cover based_on or consent_id, but the schema descriptions for those are already detailed.

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?

Opens with a specific verb and effect: 'Starts a job and reserves its credits', and the surrounding sentences (prompt_mode, model maxLength, list_models) make clear this is the model-driven image/video generation entry point. It is distinguishable from siblings like get_price, get_job, and list_models because it names them as separate steps. It stops short of explicitly naming images/videos in the description text (that lives in the title), so it is clear rather than maximally self-contained.

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 an explicit prerequisite ('Call get_price first'), explicit alternatives for receiving the result ('Pass wait_seconds to wait for the result in this call, else poll get_job', plus webhook_url in the schema), and explicit selection guidance for prompt_mode ('auto' default vs 'exact'). It also explains the required idempotency_key contract — a new key per intended job, reuse only to retry — which is exactly the kind of when-to-use rule an agent would otherwise get wrong.

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