Preview a bounded human-style input plan (move, click, drag, typing, reading) without executing it or requiring cancellation. Validate parameters and client bounds before running the real action.
Preview a bounded plan without input or cancellation. Typing previews are pure and need no hwnd; pointer previews observe explicit hwnd for client bounds. Page freshness is the caller responsibility.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Changes observed during successful MCP inspections.
v0.2.0Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false, so the description's job is the extra context, and it delivers: previews emit no input, perform no cancellation, and typing previews are pure with no hwnd. The 'page freshness is the caller responsibility' caveat is a genuinely useful disclosure that the preview does not validate live UI state, though it says nothing about what the returned preview contains or its cost.
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?
Three short sentences, front-loaded with the core distinction (preview, no input), then the per-kind hwnd rule, then the freshness caveat. Dense but every sentence carries information; no repetition of the name or of annotation-declared safety facts.
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?
This is a five-variant oneOf tool with no output schema, and the description covers the dry-run contract and hwnd asymmetry but omits the 'reading' kind entirely and never says what a successful preview returns or how that result is subsequently used. For a tool whose whole value is the preview artifact, that gap is significant.
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 high (100%), so the baseline is 3, but the description adds real meaning the schema cannot: it explains why hwnd is conditionally required — typing needs none, pointer kinds need hwnd to resolve client bounds. That rationalizes the oneOf split. It leaves the timing/tuning parameters (jitter_px, overshoot_px, pause_policy, threshold_px, wpm, typo_rate) entirely to the 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?
States a specific verb and resource ('Preview a bounded plan') and immediately narrows it with 'without input or cancellation', which separates it from executor siblings like human_type or human_mouse_click. It never names an actual alternative or explains what a 'plan' is in this system, so the agent must infer the domain from the kind enum rather than the prose.
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 phrase 'without input or cancellation' implies a dry-run to be used before committing real input, but no explicit when-to-use, when-not-to-use, or named alternative is given. An agent can guess the ordering (preview first, execute later) but the definition never confirms it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.