Skip to main content
Glama

Robot Actions — Remote Device Control

android_accessibility_audit

Run an accessibility audit on the CURRENT screen of an Android device. Checks match the rules Google's own Accessibility Scanner applies, so findings line up with what Android developers already act on. Audits whatever is in the foreground right now, so navigate to the screen you care about FIRST (device_tap_by_text / device_launch_app), then call this. Reports tap targets below the recommended minimum, several tappable elements stacked on identical bounds (the screen reader announces the same thing twice), controls that report a button class but expose no click action (the screen reader announces a button the user cannot activate), actionable controls the screen reader cannot name at all (no text, no description, no labelled child — the classic unlabelled icon button), descriptions that redundantly restate the element's type, state or action ("Submit button" — the screen reader says "button" already), and controls reporting a class assistive technology cannot interpret. Each finding carries the element's label, class and pixel rect — the rect centre is directly tappable with device_tap. Pass scroll:true to keep scrolling and re-auditing until nothing new comes into view, so content below the fold is covered too. IMPORTANT LIMITS, so findings are read correctly: tap-target results are a screening signal, not a verdict, because a control can legitimately extend its own tap area in a way this audit cannot observe; only elements actually drawn on screen are examined; and an element made tappable purely by an ancestor's touch dispatch is invisible here unless its own class is a control class. Contrast and text-size are NOT checked — that needs data this audit cannot read.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
udidYesDevice serial number (UDID)
checksNoChecks to run; omit for all. "TouchTargetSize" = tap targets below the recommended minimum; "DuplicateClickableBounds" = several tappable elements on identical bounds; "MissingClickAction" = a button-class element exposing no click action; "SpeakableTextPresent" = an actionable control with no text, content description or labelled descendant; "RedundantDescription" = description restates type/state/action; "ClassName" = control reports an uninterpretable class.
scrollNoDefaults false. When true, audit the screen, scroll down, and audit again — repeating until a pass reveals no new elements or maxScrollSteps is reached — then report the deduplicated union. This CHANGES THE SCREEN: it scrolls the app and does not scroll back, and on a lazily-loaded list it will trigger loading. Leave it off when the app must not be disturbed.
maxScrollStepsNoOnly with scroll:true. Maximum scrolls before stopping regardless (default 6, max 20).
localeIsEnglishNoDefaults true. The redundant-description check matches English words, so set false on a non-English device to skip it rather than have it compare English words against other-language text.

TDQS

A5/5.0
Behavior5/5

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

With no annotations, the description fully carries the transparency burden. It discloses that scroll:true changes the screen and does not scroll back, triggers loading on lazy lists, only examines drawn elements, cannot observe ancestor-touch-dispatched tap areas, and tap-target results are a screening signal not a verdict. It also states exactly what each finding contains (label, class, pixel rect). No behavioral surprises are left hidden.

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

Conciseness5/5

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

The description is long but every sentence earns its place. It is front-loaded with the core purpose, then logically progresses through what findings look like, how to use scroll, and important limits. The structure (purpose → usage → findings → boundaries) makes the length feel necessary and well-organized, not bloated.

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 audit tool with no output schema, the description is exceptionally complete. It describes the return value (findings with label, class, pixel rect), how to act on it (rect center is tappable), what checks are included, what limitations exist, and what is intentionally not covered (contrast, text-size). An agent can confidently decide to invoke this tool and correctly interpret its results.

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

Parameters5/5

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

Although schema coverage is 100%, the description adds substantial meaning beyond parameter names and types. It explains what each check actually detects (e.g., DuplicateClickableBounds = 'screen reader announces the same thing twice'), gives concrete examples, and explains the side effects of scroll:true and the meaning of localeIsEnglish. This significantly enriches the schema's terse enum descriptions.

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 opens with a clear verb+resource: 'Run an accessibility audit on the CURRENT screen of an Android device.' It immediately distinguishes this tool from its iOS sibling and names the device platform. It also aligns with Google's Accessibility Scanner rules, giving a concrete, recognizable scope.

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?

Explicitly states a prerequisite: navigate to the desired screen first via device_tap_by_text or device_launch_app. It also instructs when to use scroll:true ('to cover content below the fold') and when not to ('Leave it off when the app must not be disturbed'). Limitations explicitly say what this tool does NOT do (contrast/text-size), steering users away from inappropriate use.

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.

TDQS

B3.1/5.0
Disambiguation2/5

The set contains near-identical duplicate families: web_* and playwright_* expose ~15 pairs of the same desktop-grid-browser operations (web_get_text/playwright_get_text, web_reload/playwright_reload), and screenshot/log/network/mock capabilities each have 5-8 entry points (device_screenshot vs android_mjpeg_screenshot vs ios_screenshot vs ios_fast_screenshot vs web_screenshot vs webpage_screenshot vs session_screenshot). Many individual descriptions carefully draw boundaries (devtools vs traffic, HID vs session), but an agent cannot reliably distinguish web_* from playwright_*, and ios_screenshot/ios_fast_screenshot/ios_mjpeg_screenshot blur together.

Naming Consistency2/5

The prefix scheme is broken: Android functionality is split arbitrarily between android_* and device_* (device_screenshot vs android_mjpeg_screenshot), the desktop browser gets two parallel prefixes (web_* and playwright_*), and verbs vary across equivalents (device_navigate_url vs web_navigate vs ios_safari_navigate). session_* uses bare verbs (session_url, session_back), and the same concept gets different names (ios_clipboard_get_hid vs ios_get_pasteboard; device_screen vs ios_orientation).

Tool Count1/5

333 tools is an extreme count by any measure — far beyond the 50+ threshold — and much of the bulk is duplicative (the web_*/playwright_* pairs alone double ~15 slots) or out-of-scope for a device-control server (TestRail, Jira, AzDO, agent memory, secret variables, feedback). Even granting that remote device control + test automation is a broad domain, this surface will devastate agent context budgets and is impossible to navigate coherently.

Completeness4/5

The core device-control and test-automation domain is remarkably thorough: Android and iOS each have full interaction, app-lifecycle, file, network/proxy, performance, crash, accessibility, recording, and replay coverage, with CRUD lifecycles for flows, suites, app uploads, TestRail cases, and visual-review baselines. Minor gaps exist at the margins — Jira/AzDO lack update/transition/comment operations, and iOS cannot open/close tabs — but the central workflows have no dead ends.

Resources