find

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

Beyond the annotations (readOnly=false, destructive=false, openWorld=false), the description discloses latency (about 0.3s vs 2–5s for look), that look mode two-passes the screenshot (coarse window box, then zoomed detail), that screenshots are sent only to Jev and not into the caller's context, and that look requires an official key (status jev_images=true). It leaves open what is returned on failure or when confidence falls below min_confidence.

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 core capability is front-loaded in the first clause, followed by the fast path, the visual path, and the fallback in a logical escalation order. It is a dense run-on paragraph with several chained qualifiers, but almost every clause carries operational 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 9-parameter tool with no output schema, the description explains the main modes and confidence semantics but does not say what the response contains (element handle, coordinates, ranked candidates, failure shape) or what the remaining parameters do. An agent can invoke it but may be unsure how to consume or retry the result.

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?

At 67% schema description coverage the description still carries extra meaning: it explains the click=true behavior and its confidence gate, the semantics of confidence as a post-zoom probability for look mode, and region's role as a look-mode search restriction. The `count` and `button` parameters remain undocumented in both description and schema.

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: describe the target in one sentence and Jev selects the matching UI element and returns a confidence. It also distinguishes the fast text/tree path from the visual path and names `screenshot` as an escape hatch. It does not distinguish itself from the sibling `click` or `observe` tools, which overlap in intent.

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?

Clear routing rules are given for the modes: default text/name matching, add look=true for icons/game visuals/images/colors, escalate to `screenshot` if that fails. What is missing is guidance on when to call `find` versus the sibling `click` tool, given that click=true makes this tool perform clicks directly.

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