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

Annotations already establish readOnlyHint=true, idempotentHint=true, and destructiveHint=false, so the safety profile is known. The description adds valuable behavioral context beyond annotations: platform-specific scoring thresholds (44pt vs 48dp), Android warning relaxation ('onSurfaceVariant/outline = warn not fail'), the hosted-endpoint limitation on taste-profile bindings, and that the screenshot is only for the caller's reference since 'Geometry is scored from the snapshot, not decoded pixels.' No contradiction with annotations.

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?

Seven sentences, each earning its place: purpose, self-discovery hint, platform behavior, scoring scope, return reference, endpoint limitation, and stateless usage. Front-loaded with purpose, no filler or redundancy with the schema fields. This is dense but efficiently structured.

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 7-parameter tool with nested objects (elements, viewport) and no output schema, the description covers purpose, platform behavior, scoring scope, return shape (delegated to audit_page reference), and a key limitation. The 'Same return shape as audit_page' shortcut assumes agent knowledge of that sibling, but the self-discovery instruction ('Call with no arguments') provides a fallback for both input shape and capture method. Reasonably complete for the complexity.

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?

With 86% schema coverage the baseline is 3. The description goes beyond the schema by explaining the behavioral impact of platform (threshold and semantic-role differences), the meaning of omitting project/profile (stateless run; profile bindings unavailable remotely), and the self-documenting hint 'Call with no arguments for the expected snapshot shape' for elements. This adds genuine semantic value the schema's field descriptions don't provide.

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 first sentence states a specific verb-resource pair: 'Audit a rendered mobile screen (iOS or Android) from a view-hierarchy/accessibility snapshot.' It further disambiguates from the many audit_* siblings by scoping exactly what is scored ('touch targets, contrast, and visual rhythm') and by referencing the analogous 'Same return shape as audit_page,' making the tool's distinct role clear.

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 actionable invocation guidance: 'Call with no arguments for the expected snapshot shape,' explains when to pass platform:"android" vs the iOS default, and notes 'Omit project/profile to run the stateless screen audit.' It does not explicitly name alternatives or state when NOT to use this tool over sibling audit_* tools, but the platform and binding trade-offs are well covered.

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