lamda

Supports Cursor, Mcp

FIRERPA provides a browser-based remote desktop: view and control the device in real time without installing a dedicated PC client, over LAN or across networks. Streaming supports both MJPEG and H.264, with software and hardware encoding backends. Frame rate (up to 60fps), resolution scale, bitrate, and quality can be tuned to device capability and network conditions—reducing bandwidth and improving smoothness on weak links. WebRTC is also supported, with configurable STUN/TURN servers for better NAT traversal and lower end-to-end latency in public or cross-region setups.

The remote desktop supports multi-user concurrent access for collaborative debugging, demos, and training. Bidirectional clipboard sharing and live audio (Android 10+) are included, along with an integrated terminal, drag-and-drop uploads, and browser-based file browsing and downloads—all over a single port. Built-in visual layout inspection highlights elements, supports Tab traversal, shows coordinates and RGB values, and exports the XML layout tree so you can validate selectors and automation logic in the same UI, shortening the loop from screen to script.

Remote desktop and RPC support end-to-end TLS and service-certificate access control, with optional custom WebUI login passwords to reduce exposure on public networks. Remote desktop capabilities can also be embedded into your own web apps over WebSocket (live video, touch, terminal, keys, etc.), with allow_origin for cross-origin integration—suitable for productizing real-device control.

FIRERPA provides a full selector-based automation system: text, resourceId, description, scrollable, and other common matchers, plus child / sibling chaining for duplicate elements, deep hierarchies, or controls without distinctive attributes. At the element level: screenshots, wait for appear/disappear, corner/center coordinates, exists checks, Unicode input, stepped swipes, fling scrolling, scroll-to-end, and more. It can coexist with other accessibility services on Android 8.0+ and improves WebView node discovery for hybrid apps.

The UI Watcher listens for UI changes in real time and runs clicks, key events, or counters when conditions match—useful for auto-dismissing agreements, update prompts, ads, and other interruptions. Multiple selector conditions and per-event enable/disable are supported; transient screens can be counted.

Virtual Display is a differentiator: create isolated background displays on the device and run apps and automation there without affecting the main screen—e.g. auto-reply on a virtual display while you use the phone normally. The virtual-display API mirrors the main device API (d.xxx → vd.xxx); Watchers can be scoped to a virtual display. The WebUI supports multi-display view and switching in remote desktop—ideal for background tasks, parallel work, and human-in-the-loop plus machine automation.

For scenes without a standard view tree (games, custom-drawn UI), FIRERPA offers OCR and image matching. OCR supports paddleocr, easyocr, and custom HTTP backends; in clusters, recognition can run centrally instead of loading models on every PC, with GPU acceleration and flexible text matching. Image matching runs on the device (template or SIFT), without consuming PC resources; SIFT is robust to rotation, scale, and lighting. Multi-touch supports recording, replay, programmatic construction, and binary persistence for complex gestures and pressure.

Architecturally, FIRERPA uses a client/server model—better for centralized scheduling, versioning, and fleet control than on-device script runners like AutoJS; lighter than Appium; more stable than uiautomator2 in multi-device scenarios. The docs position it as a functional superset of common Android automation stacks—automation, capture, Hook, and ops can be chained in code on one platform without switching tools.