Skip to main content
Glama

Robot Actions — Remote Device Control

iOS VoiceOver

ios_voiceover

Drive iOS VoiceOver on a real device and read where its cursor lands. Actions: 'status' (is it on), 'start' / 'stop' (turn VoiceOver on or off — a few seconds each; 'enable' / 'disable' are accepted as the same, matching android_talkback), 'next' / 'previous' (move to the next or previous element), 'activate' (VoiceOver's double tap on the current element), 'back', 'home', 'rotor' (change the rotor setting — Headings, Links, Words…; pass direction), 'rotor_item' (previous / next item of the current rotor setting; pass direction), 'read_all' (read from the top for duration seconds, then pause). Each move returns text — the accessibility label of the element VoiceOver settled on, plus its focus (label, frame in points, app). The exact spoken sentence (roles, hints) is redacted by iOS; the label is what is available — an element VoiceOver reports without a label (some search fields) shows as "(no label)" even if its page-source entry has a label or placeholder, so check ios_page_source before calling it unlabelled. text is null when VoiceOver did not report a new position (end of the screen, or an activate — an activate usually works but the new position is not reported until the next move, and its note says so); right after start there is no cursor yet, so a first move that lands nowhere is retried once and then carries a note — move once more. read_all reads at speaking pace, a few seconds per item, so its list holds only what was reached in duration seconds (the reply echoes the duration used); the list can miss the first item, so use next / previous for a complete transcript; its final position is the last item listed. If a move reports no position and VoiceOver has switched itself off (a system prompt it cannot reach may be showing), the reply says enabled: false. On a device seeing VoiceOver for the first time, iOS shows a one-time "VoiceOver changes the gestures" sheet that holds the cursor until someone answers Use VoiceOver. While VoiceOver is on, taps and swipes behave differently (one tap selects, a double tap activates), so stop it when done. The device is kept silent automatically while VoiceOver is on, unless someone is watching it in a browser, who hears it instead.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
udidYesDevice UDID
actionYesWhat to do. 'status' changes nothing.
durationNoFor 'read_all' only: seconds to read for. Default 30.
directionNoFor 'rotor' and 'rotor_item' only. Default 'next'.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • changedInput schema / properties / duration / description
      Previous value: -"For 'read_all' only: seconds to read for. Default 10."New value: +"For 'read_all' only: seconds to read for. Default 30."
  2. Changed1 schema field changed
    • changedInput schema / properties / action / enum
      Previous value: -[
      -  "status",
      -  "start",
      -  "stop",
      -  "next",
      -  "previous",
      -  "activate",
      -  "back",
      -  "home",
      -  "rotor",
      -  "rotor_item",
      -  "read_all"
      -]New value: +[
      +  "status",
      +  "start",
      +  "stop",
      +  "enable",
      +  "disable",
      +  "next",
      +  "previous",
      +  "activate",
      +  "back",
      +  "home",
      +  "rotor",
      +  "rotor_item",
      +  "read_all"
      +]
  3. Added

TDQS

A4.8/5.0
Behavior5/5

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

Far exceeds the annotations (which only state readOnly=false, destructive=false, idempotent=false): it discloses the one-time 'VoiceOver changes the gestures' sheet blocking the cursor, null `text` on end-of-screen or activate, retry-then-note behavior after start, silent-device behavior, and enabled:false when a system prompt intervenes.

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?

Front-loaded with the action inventory, then caveats, so an agent gets the essentials first. It is dense and long, but nearly every sentence carries operational detail; only minor tightening is possible.

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?

For a complex multi-action tool with no output schema, it fully describes return shape (`text`, `focus` with label/frame/app, notes), failure modes, and prerequisite state. Nothing needed to invoke it correctly is missing.

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 already 100%, so the schema documents udid/action/duration/direction. The description adds real semantic value over baseline by explaining each action's effect, that 'status' changes nothing, that duration governs read_all pacing, and that direction applies to rotor/rotor_item.

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?

States a specific verb+resource ('Drive iOS VoiceOver on a real device and read where its cursor lands') and enumerates every action with its meaning. It explicitly distinguishes itself from siblings by noting 'enable'/'disable' mirror android_talkback and by routing unlabelled-element checks to ios_page_source.

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

Usage Guidelines5/5

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

Gives explicit when-to-use guidance: use next/previous for a complete transcript instead of read_all, check ios_page_source before calling an element unlabelled, move once more after a start retry, and stop VoiceOver when done because taps/swipes behave differently.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources