agent-device

agent-device is a device automation CLI and integration layer that provides AI agents with mobile and desktop app verification.

Agent-device provides device automation for AI agents operating on iOS, Android, TV, and desktop environments. It gives agents a structured way to understand screens by exposing accessibility trees with stable references, keeping context smaller than raw screenshots. Agents can act using semantic selectors to find elements by text, label, or role. The platform supports development simulators, physical QA devices, and TV-class targets with a single workflow. It includes a typed TypeScript client, CLI, and official skills for integration into codebases and custom cloud agents. Users can capture parsed logs, network dumps, and screen recordings for debugging and evidence.

agent-device at a glance

Best for: Software engineer
Runs on: CLI, API, macOS, iOS, Android, Linux
Pricing: Free
Released: January 2026
Category: AI coding tool
Website: agent-device.dev

Key features

Token-efficient UI snapshots
Stable element references
Semantic selectors by text, label, or role
Cross-platform support for iOS, Android, and TV
Exploration to E2E script replay
Typed TypeScript client and CLI
Parsed log and network dump capture
Screen recording and visual proof

Use it for

Exploratory QA on mobile and desktop apps
Bug reproduction with captured logs and steps
Regression loops using replayable scripts
Clean-state testing and permission management

Alternatives to agent-device

Agent OS — Manage AI coding sessions side by side and check on them from your phone.
Mobilerun — Control real mobile devices and automate tasks with any large language model.
Composio — Equips AI agents with tools, managed user auth, and sandboxes across 1,500+ apps
AG-UI — Standardizes agent-user interactions and generative UI for applications.
sim-use — Provides token-efficient screen outlines and alias-cached taps for fast agent loops.
AgentDock — Manages multiple coding agents from one screen-edge inbox without switching tabs.
PhoneClaw — It executes AI agent tasks and system tool calls entirely on-device.
Mobile MCP — Connects AI agents directly to real mobile devices for automated testing and tasks.
AgentDock — It coordinates multi-device workflows and provides secure environment access.
iphone-use — Operates real iPhones via HTTP and MCP without touching the host screen.
ARTEMIS — Connects AI agents to real Android devices for autonomous UI testing.
Agent QA — It uses repository-based memory to learn from test runs and perform self-healing.

All agent-device alternatives