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

Annotations already declare readOnlyHint=false and openWorldHint=true, so the non-read-only nature is covered. The description adds substantial value beyond annotations: deferred settlement (charged only on completion, never for failures), asynchronous behavior with job_id and 5-15 min runtime, and the idempotency_key dedup mechanism. The idempotentHint=false annotation does not contradict the idempotency_key discussion — the description explains how to achieve idempotent replay, not that the operation is inherently idempotent. Minor demerit for repeating the 'never charged' guarantee twice.

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 core content is front-loaded with the paid status and pipeline, which is good. However, the bracketed 'Paid:' tail repeats payment and settlement facts already stated in the opening ('5 USDC', 'never charged', deferral, polling) nearly verbatim. This duplication inflates the description and adds noise. The idempotency section is also stated twice (in prose and again in the schema's idempotency_key description), making the overall structure less tight than it could be.

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 complex paid, asynchronous tool with side effects, the description covers the essentials: cost, settlement conditions, polling mechanism, runtime estimate, and retry/idempotency safety. There is no output schema, so return-value documentation falls to the description; it does mention job_id is returned. It could note error/success payload shapes or edge cases beyond 'failed generation', but the operational envelope an agent needs to call this safely is well covered.

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% and every parameter (prompt, platform, asset_kind, polygon_tier, texture_size, idempotency_key) already carries a meaningful description in the schema, including enums and defaults. The tool description adds little beyond what the schema states — e.g., it reiterates the idempotency_key retry semantics that the schema already documents. Baseline 3 is correct since the schema does the heavy lifting.

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 states a specific verb and resource — 'generate a production 3D avatar from a text prompt' — and specifies the pipeline (xAI concept → TRELLIS.2 mesh → post-process). The phrase 'from a text prompt' implicitly distinguishes it from the sibling generate_avatar_from_image, and the paid/async framing separates it from get_generation_status. An agent can tell exactly what this tool does and how it differs from its nearest sibling.

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 operational usage: how to retrieve results ('poll with the free get_generation_status tool'), when retries are appropriate ('after a timeout/5xx'), and how to guard them ('Pass idempotency_key'). It also signals cost-vs-alternative tradeoffs (paid vs free polling). It does not explicitly name when NOT to use this tool versus generate_avatar_from_image (e.g., when an input image already exists), which would push it to a 5.

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