preview

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

With no annotations at all, the description carries the full burden, and it does disclose genuinely non-obvious behavior: asynchronous execution, single most-recent-job retention, eight retained PNG cache files, and id invalidation on revision change. However it never explains how the async result is retrieved or when the job is complete, says nothing about side effects on disk or project state, and gives no error or permission context. Useful behavioral signal, but the async contract and side effects are underspecified.

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 text is dense and mostly earns its length given 9 parameters and multiple modes, but it is not front-loaded: it opens mid-thought with "Default native renders..." rather than stating what the tool does. Sub-modes run together in prose without structure, and terms like "without export or focus" are unexplained, slowing scanning for a long description.

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 tool with a nested required `target` object, an unknown `observation_token` credential, no annotations and no output schema, the description leaves too much undone. It explains the mode nuances but not the two required inputs, not the response shape, and not how an async preview is obtained — the agent cannot confidently construct a call from this alone.

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% across 9 parameters, so the description must compensate and it only partly does: it defines `kind` values, the `max_size` range (128..1600), `points` as {x,y} pixel-edge coordinates, `preview_id` reuse/invalidation, and implies `direction` by explaining the RAW-to-normalized division. Both required parameters, `target` (a nested open object) and `observation_token`, are never mentioned, nor is the `action` enum's full set (prepare, restore_destination). Two undocumented required params is a serious gap.

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 does convey a verb and resource ("renders current C1 adjustments asynchronously") and enumerates the kind modes, so an agent can infer this produces previews of a photo's edits. But it never states the purpose in a plain leading sentence and leans on undefined jargon ("C1", "native", "focus"), and it draws no distinction from siblings like `images` or `read`. The tool name "preview" plus mode-specific detail leaves the core purpose only implicit.

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 routes between modes explicitly: `kind=mask` for the selected layer's composite mask, `kind=render` as legacy export, `kind=viewer/action=capture` to sample the display, and `map_points` must reuse the same kind/max_size and returned preview_id. It also states invalidation conditions (changed photo, layer or revision) and retention behavior, which tells the agent when a cached id is still usable. What's missing is guidance on the non-legacy `action` values and when to prefer this tool over sibling tools.

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