Argent CLI
Gives an agent control of iOS simulators, Android emulators, TVs and Electron apps: taps, flows, profiling and React Native debugging
- 3,123 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- The assistant taps and types inside a running app and on a device
- Debugging reads app logs and network requests and evaluates JavaScript in the app
- The installer edits the editor's MCP configuration and permissions, and telemetry is on by default
Install
- Works in
Manual install
npx @swmansion/argent@latest initFor pnpm: pnpm dlx @swmansion/argent@latest init. The wizard detects the editor and writes MCP configuration.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The assistant taps and types inside a running app and on a device
- Debugging reads app logs and network requests and evaluates JavaScript in the app
- The installer edits the editor's MCP configuration and permissions, and telemetry is on by default
This is third-party code. Review the repository files before installing.
What it does
Argent connects an AI assistant to a running app: on an iOS simulator, an Android emulator or device, on Apple TV, Android TV and Fire TV, and in Electron and Chromium. The assistant taps, swipes, types and presses hardware buttons; TVs get a D-pad. Flows are recorded once and replayed as a bug repro or smoke test. A visual check diffs a baseline against a live capture with OCR and font awareness. Profiling hooks into Hermes, React DevTools, Xcode Instruments and Perfetto: renders, CPU hotspots and hangs. Debugging covers logs, network requests, JS evaluation and native and React trees.
Who it is for. For React Native and native mobile developers who want the agent to drive the app, reproduce bugs and capture profiles.
Good fit when
- A bug reproduces only after a chain of actions in the app and you need it recorded as a flow
- The app is slow and you need a profile of renders and CPU hotspots without hand work in Instruments
- You need to compare a screen against a baseline capture after a layout change
- You need a smoke test on a simulator that the agent runs after changes
- The project is React Native and the agent should build, launch and iterate with no extra setup
Not a fit when
- The app is a web page in a regular browser: browser MCP servers fit that
- You need unit tests that do not launch the app, while Argent drives a live app
- Telemetry is not allowed by company policy and you will not disable it
- You only need basic Android or iOS control and a simpler MCP is enough
Example request
Launch the app on the iOS simulator, walk through the checkout screen and capture a render profileLimitations
The license is mixed: source code is Apache 2.0, while the simulator-server and ax-service binaries and the .dylib files belong to Software Mansion and are licensed only for use within this project. Telemetry is on by default and turns off with argent telemetry disable. The iOS side needs macOS with Xcode and simulators. Supported platforms and editors are listed in the docs at docs.swmansion.com/argent.
How to disable. The repository ships an uninstall command: use the Argent installer's uninstall and remove Argent entries from the editor's MCP configuration. Telemetry turns off with argent telemetry disable.
FAQ
How do I install Argent for Claude Code or Cursor?
Run npx @swmansion/argent@latest init, or pnpm dlx @swmansion/argent@latest init. The wizard writes MCP configuration for the editors it finds.
What are flows in Argent?
A path through the app is recorded once and replayed deterministically as a bug repro or smoke test.
Which profilers does Argent connect?
Hermes, React DevTools, Xcode Instruments and Perfetto. They show renders, CPU hotspots and hangs.
Does Argent work with Apple TV and Android TV?
Yes, Apple TV, Android TV and Fire TV are supported with D-pad control. Electron and Chromium work with mouse and keyboard.
Does Argent collect telemetry?
Yes, by default. Turn it off with argent telemetry disable. The list of collected data is described in the repository's Telemetry.md.
Discussion
No questions or reviews yet.
Related
CLI toolsThe npx serve-sim command streams an iOS Simulator to the browser and lets an agent tap, swipe, type and replace the camera feed
Buoy MCP lets an agent read a running React Native or Flutter app, tap its screen and inspect the network
A set of skills, agents and commands for iOS, iPadOS, watchOS and tvOS development: Swift 6, SwiftUI, Liquid Glass and Apple Intelligence
Lets Claude Code and Codex drive a real iPhone: read the screen, tap, type and scroll in any app
Screenshots and preview videos for the App Store and Google Play: the agent walks the app on a simulator, frames the shots and checks store requirements
An MCP server to drive iOS and Android: simulators, emulators and real devices via accessibility trees and screenshots
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Argent (this entry) | no | 3,123 | ||
| Mobile MCP | no | not required | 10,095 | Basic device control; Argent adds flows, profiling and React Native debugging |
| iOS Simulator MCP Server | no | not required | 2,201 | iOS simulator only, with no Android or profilers |
| Debroid | no | 276 | a headless Android debugger on JDWP |
README in short
The description opens with the promise that the assistant drives the app without leaving the CLI. Six capabilities follow: interaction, flows, visual regression, profiling, debugging and React Native. Links lead to installation, platforms, CLI and editors. It closes with the opt-out telemetry, the mixed license and a note that Software Mansion built Argent.