Argent

Argent lets an AI coding agent run, inspect, debug, and profile apps on its own, so it can see what happened on screen and keep working in the same session. It reaches iOS Simulators, Android emulators and physical devices, TVs (Apple TV, Android TV, and Fire TV), and Electron or Chromium desktop and web apps. At its core it is a set of agent skills plus an MCP server that connects coding harnesses to those targets.

For interaction, the agent taps, swipes, types, presses hardware buttons, uses a D-pad on TVs, and drives mouse and keyboard on desktop, navigating by coordinates from accessibility trees. Multi-step interaction sequences run in a single call, and Argent reports back after every interaction. Paths can be recorded once and replayed deterministically as a reproduction or smoke test, and visual regression diffs a baseline against a live capture with OCR and font awareness.

For debugging, it attaches a debugger, reads console logs, evaluates JavaScript expressions, explores the native view hierarchy and the React component tree, and reads network requests and payloads at both the JavaScript fetch layer and the native network layer. For performance, it records React and native iOS profiles together through Hermes, React DevTools, Xcode Instruments, and Perfetto, traces slow React commits down to native stack frames, and detects UI hangs, render cascades, and memory leaks. React Native apps can be built, launched, and iterated on without extra setup.