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

Beyond the annotations it discloses that fixes are written "as a new version" (consistent with destructiveHint=false), that most modes and every fix consume credits, and that persona is free — non-obvious cost and side-effect context an agent can't get elsewhere. It does not explain the confirmed/confirmationToken consent handshake or whether runs are synchronous, so the mutation/consent profile is only partly covered.

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?

It is front-loaded with the verb and organized mode-by-mode, roughly 120 words with no filler sentences. It is compressed to the edge of cryptic in places ("fix only, no audit: apply: true with the violations you have"), 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 five-mode tool with no output schema, return values are only sketched ("issues by severity, accessibility included"), leaving what heatmap, gaze_flow and a fix run hand back unclear, and the consent flow implied by confirmed/confirmationToken is never mentioned in the description. The input side is well covered; the outcome and workflow side is thin.

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?

Because params is an untyped nested object (additionalProperties), the description carries most of the real semantics by naming designId, html, url, image, imageUrl and pageId, and by noting image/imageUrl apply only to heatmap and gaze_flow and pageId targets a specific canvas page. That is genuine added meaning over the schema, though confirmed/confirmationToken are left entirely to the schema text.

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 opens with a specific verb+resource ("Review a design") and immediately disambiguates the five modes it covers (ux, persona, heatmap, gaze_flow, accessibility) with a one-phrase gloss each. An agent can tell exactly which capability this tool exposes and what each mode produces without opening the 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?

It gives real selection guidance: accessibility is "fix only, no audit" and needs apply:true plus violations you already have, while apply:true works on ux/persona output; it also flags that persona is free and everything else spends credits, which affects the decision to call. What's missing is any comparison to sibling tools (read, design, edit) or a statement of when not to use review at all.

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