look

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

Annotations already cover readOnly/idempotent/non-destructive, so the description earns credit for adding behavior beyond them: it operates on covered/background windows without raising them, falls back to an image only if the tree is too thin, and notably does not re-attach the image if the screen matches the last one sent. That image-dedup rule is genuine behavioral context an agent could not infer from structured fields.

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 critical rule is front-loaded and every clause carries information, but the prose is over-compressed into semicolon run-ons and fragments ('Unlocks a batch.'), which costs parseability for an agent reading the definition cold. It earns its length but not its density.

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, zero-annotation-coverage tool with no output schema, the description is largely sufficient: it explains the return format (one line per element, image fallback, compact JSON shape) and the main parameter semantics. The unexplained max_depth/max_elements/output_path leave a residual gap that keeps it from a 5.

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 0%, so the description carries the full burden and largely does: hwnd accepts an id, a window title, or a process name (much richer than the bare anyOf), query filters, format="json" yields compact JSON with box_2d in 0-1000, annotate draws numbers, zoom=box_2d enlarges an area, capture_screenshot forces the image, WebP capped at 2000px. max_depth, max_elements, and output_path are never explained, which keeps it off a 5.

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 concrete verb and resource: see a NAMED window (hwnd, title, or process) without bringing it to the front, emitting one line per element with role/name/value/states/coordinates. It is specific enough to distinguish from 'act' (which it feeds) and from 'windows' (which presumably lists rather than inspects), though it never explicitly contrasts with 'window' or 'windows'.

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 front-loaded 'GOLDEN RULE: look → act in one batch, targets by label; never resend a batch without looking' gives an explicit workflow and names the sibling it precedes ('act'), and 'Unlocks a batch' reinforces the sequencing. What is missing is a when-not condition or guidance on choosing 'look' over the 'window'/'windows' siblings.

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