Robot Actions — Remote Device Control

Webpage type

webpage_type

Type text into an input/textarea in the device browser, by CSS selector or by a ref from a page-elements listing — works on BOTH iOS Safari and Android Chrome. On iOS it taps the field and types on the DEVICE keyboard, so the page gets real trusted keystrokes (keydown, keypress, beforeinput, input) exactly as from a person — use this for anything that reacts to typing rather than just reading .value. Otherwise it sets the value via the native HTMLInputElement/HTMLTextAreaElement value setter (so React/Vue-controlled inputs register the change) and dispatches bubbling input and change events, which are isTrusted:false. The response reports which happened via via ("nativeKeyboard" or "jsValue"), and when it fell back to jsValue it names the cause in fallbackReason — you never have to guess whether a degraded path was taken. Pass require:"native" to make an unavailable device keyboard a hard error instead of a silent downgrade, for fields that only behave correctly under real keystrokes. exact:false means the keystrokes landed but the component rewrote the value as you typed (masks, autocompletes) — that is a success, not a failure. By default appends to the existing value; pass clear:true to replace it.

Input Schema

TableJSON Schema

Name	Required	Description
`ref`	No	Element ref from a page-elements listing, e.g. "e12". Provide this or `selector`. A ref points straight at the element it was issued for, so it survives markup that shifted position, and it can address elements no CSS selector can reach from the top of the page.
`text`	Yes	Text to type
`udid`	Yes	Device UDID / serial (iOS or Android)
`clear`	No	Replace the existing value instead of appending (default: false)
`pageId`	No	Target page/tab id — auto-picked when omitted
`socket`	No	Android only: abstract unix socket name (default: chrome_devtools_remote)
`require`	No	Set to "native" to fail loudly when the device keyboard is unavailable, instead of silently falling back to setting .value (isTrusted:false). Use for fields that gate on real keystrokes. iOS only — Android has no native-keyboard path here.
`selector`	No	CSS selector for the target element. Provide this or `ref`.
`platformVersion`	No	Ignored (kept for compatibility)

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedAug 18, 2026

addedInput schema / properties / require

Added value: +{
+  "description": "Set to \"native\" to fail loudly when the device keyboard is unavailable, instead of silently falling back to setting .value (isTrusted:false). Use for fields that gate on real keystrokes. iOS only — Android has no native-keyboard path here.",
+  "enum": [
+    "native"
+  ],
+  "type": "string"
+}

Changed3 schema fields changedAug 9, 2026

addedInput schema / properties / ref

Added value: +{
+  "description": "Element ref from a page-elements listing, e.g. \"e12\". Provide this or `selector`. A ref points straight at the element it was issued for, so it survives markup that shifted position, and it can address elements no CSS selector can reach from the top of the page.",
+  "type": "string"
+}

changedInput schema / properties / selector / description

Previous value: -"CSS selector for the input/textarea element"New value: +"CSS selector for the target element. Provide this or `ref`."

changedInput schema / required

Previous value: -[
-  "udid",
-  "selector",
-  "text"
-]New value: +[
+  "udid",
+  "text"
+]

First observedAug 4, 2026

TDQS

A4.6/5.0

Behavior5/5

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

Beyond the annotations (which only flag it as a non-readonly, non-idempotent mutation), the description discloses the two execution paths, the actual DOM events dispatched (keydown/keypress/beforeinput/input vs. bubbling input/change), that jsValue events are isTrusted:false, that default behavior appends rather than replaces, and how degradation is reported. This is unusually rich behavioral disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The core action and target are front-loaded in the first clause, and the paragraph is dense with no filler sentences. It is long for a single unstructured block, though, and would scan faster if the fallback/reporting details were broken out.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

There is no output schema, and the description compensates by explaining the response shape (via, fallbackReason) and error modes. Combined with annotations covering the safety profile and a fully described 9-parameter schema, an agent has everything needed to call this correctly and interpret the result.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema baseline is 3, but the description adds real semantics: clear:true replaces vs. default append, require:"native" turns a soft downgrade into a hard error, and exact:false is reframed as success rather than failure. It also documents response fields (via, fallbackReason) that no schema declares.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description names a specific verb+resource ('type text into an input/textarea in the device browser') and immediately scopes it by target ('by CSS selector or by a `ref`') and platform ('BOTH iOS Safari and Android Chrome'). It is clearly distinguishable from siblings like webpage_press_key, webpage_click, or webpage_fill_form.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives a clear use condition: 'use this for anything that reacts to typing rather than just reading .value,' and explains when to escalate with require:"native" for fields that gate on real keystrokes. However, it never names or contrasts an alternative tool (e.g., webpage_fill_form or device_type), so the agent has no explicit routing guidance against siblings.

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

All Robot Actions — Remote Device Control tools Inspect

Try in Browser

Give agents eyes on the webHosted screenshot MCP. Send a URL, get a pixel-perfect image. $0.002 per grab. Works with Claude, Cursor, and Codex.grabbit.liveAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?