agent-device
CLI, MCP server and typed Node.js API so an agent can verify its changes in a live mobile app on iOS, Android and more
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Launches and controls apps on devices, emulators and simulators
- Reads screen state, logs and network data from the app
- Can connect to remote device clouds
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/agent-deviceDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Assembled automatically, review before installing.
Install the tool
npm install -g agent-deviceInstall the CLI: npm install -g agent-device@latest, then check the environment with agent-device doctor. For MCP tools run agent-device mcp and add the server to your agent's configuration.
Other ways from the author
npm install -g agent-device@latest
agent-device doctorRun doctor before handing the CLI to an agent to check the platform environment.
This is third-party code. Review the repository files before installing.
What it does
The tool gives a coding agent a feedback loop from a running app: it inspects the screen, controls it, captures diagnostics and verifies the result. The agent reads token-efficient accessibility-tree snapshots instead of reasoning over screenshots alone, acts through element refs and selectors, and saves screenshots, video and logs as evidence. It supports iOS, Android and HarmonyOS on simulators, emulators and physical devices, plus tvOS, Android TV, Vega OS, web, macOS and Linux. You can drive it through the CLI, the built-in stdio MCP server or a typed Node.js API, so any agent that runs commands or speaks MCP can connect. Working steps can be saved as .ad scripts for reuse locally and in CI.
Who it is for. For mobile developers and QA engineers who need an agent to verify the app on a device.
Good fit when
- You want the agent to verify its change in a running mobile app
- You need bug evidence: screenshots, video, logs and traces
- You want to repeat scenarios on devices locally or in CI
Not a fit when
- The target platform environment is not set up: Xcode and simulators, Android SDK and emulators
- You need manual acceptance rather than automated agent verification
Example request
Open the app on the iOS simulator, run the login flow and save a screenshot and logs as evidenceLimitations
It requires Node.js 22.12 or newer, 24 or newer for web automation, and a set-up target-platform environment: Xcode and simulators for iOS, Android SDK and emulators for Android, and so on. Check readiness with agent-device doctor. Some features and the command set depend on the platform and installed tools. Reliable snapshots need clear labels, roles and test IDs in the app's accessibility tree.
How to disable. Remove the global package: npm uninstall -g agent-device. Remove the agent-device block from your MCP configuration if you added it.
Security check
- Launches and controls apps on devices, emulators and simulators
- Reads screen state, logs and network data from the app
- Can connect to remote device clouds
README in short
The README describes a tool for automating and verifying mobile apps with coding agents through a CLI, a built-in MCP server or a typed Node.js API. The agent reads accessibility-tree snapshots, acts through element refs and selectors, and saves screenshots, video and logs as evidence. The quick start shows npm installation, the doctor command and an example flow in the iOS Contacts app. It works with Claude Code, Codex, Cursor, Windsurf, Cline, Goose and any agent that can run a CLI or speak MCP. MIT licensed.
FAQ
Is it a CLI or an MCP server?
Both. There is a global CLI, a built-in stdio MCP server via agent-device mcp and a typed Node.js API, all three over the same execution path.
Which platforms are supported?
iOS, Android and HarmonyOS on simulators, emulators and physical devices, plus tvOS, Android TV, Vega OS, web, macOS and Linux.
Related
A CLI for testing and red teaming LLM applications: compares models, runs automated CI checks and finds vulnerabilities
Playwright CLI
playwright-cli
Microsoft's official Playwright CLI with an agent skill: drive a browser through short commands without heavy MCP schemas
The official MCP server debugger: web UI, automation CLI and terminal UI in one package
SonarSource's official MCP server: quality and security issues, quality gates and code analysis from SonarQube Server and Cloud