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

Annotations mark this non-read-only and non-idempotent, and the description earns its keep by resolving the apparent tension: a 'preview' that nonetheless mutates nothing until confirm_change is called. That two-phase, no-side-effect-until-confirmed behavior is real context beyond the annotation flags, though permissions, rate limits, and what the preview returns go unstated.

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?

Four sentences, front-loaded with the core action and closed with the critical safety constraint. The clause 'with the prompt runs it can spend and its maximum provider attempts' is somewhat convoluted but stays within reasonable length.

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 preview tool with no output schema and four optional parameters, the description covers the essential flow and the confirm gate. It falls short on what the preview actually returns (runs, spend estimate) and leaves several optional parameters unexplained, but nothing critical to a correct two-step call 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 description coverage is 0%, so the description carries the param burden, yet it explains only market_ids ('null is the default market') and max_attempts ('cap the attempts'). prompt_ids vs prompt_set_id is loosely implied by 'a prompt set (or chosen prompts),' while repetitions, engines and project_id remain undocumented—a significant gap at seven parameters even allowing for self-evident enums.

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 precise verb and resource: 'Preview launching an audit of a prompt set (or chosen prompts) on the given engines.' It also positions itself against the sibling confirm_change, so the agent can tell the two-step pair apart without inspecting either 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?

Explicitly describes the workflow: this is a preview, 'Nothing runs until the user agrees and you call confirm_change.' It also routes per-market behavior to market_ids and hints the ids come 'from the project context.' No explicit when-not-to-use case, but the sequencing with confirm_change covers the main decision.

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