# lamda [Health: Active]

**Category:** 🖥️ OS Automation  
**Repository:** https://github.com/firerpa/lamda  
**GitHub Stars:** 8760  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/lamda

## Description
Android Full-Stack Device Control Platform: WebRTC/H.264 remote desktop, UI/OCR/image-matching automation, one-click MITM, built-in Frida, proxy/VPN/frp/P2P networking, MCP/Agent, 160+ APIs, designed for multi-device clusters and engineered deployments.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What lamda does

lamda exposes FIRERPA’s Android device-control platform to MCP-based agent workflows. It combines selector-based UI automation with OCR, image matching, screenshots, gestures, UI watchers, remote desktop access, packet capture, Frida instrumentation, terminal operations, and device networking.

The device-side service can operate without additional runtime dependencies and supports Android 6.0 and newer. Root access is optional. A PC-side Python client can coordinate multiple devices and chain actions from automation, inspection, traffic analysis, reverse engineering, and operations workflows.

## How it works

FIRERPA uses a client/server design. The Android device hosts the service, and the client communicates with it through the platform’s APIs. This arrangement supports centralized scheduling, fleet control, and shared configuration rather than requiring every automation script to run locally on the phone.

Agents can inspect the interface through selectors such as text, resource IDs, descriptions, and scrollable properties. Child and sibling selector chaining helps address repeated or deeply nested controls. For custom-drawn interfaces and games, OCR and image matching provide alternatives to the standard Android view hierarchy. OCR can use PaddleOCR, EasyOCR, or a custom HTTP backend; image matching runs on the device using templates or SIFT.

Virtual displays allow applications and automation tasks to run in the background without taking over the main display. The same device API is mirrored for virtual displays, and watchers can be assigned to a specific display.

## Setup and configuration

Run the FIRERPA service directly on an Android device, then connect through the PC-side Python client or MCP integration. The device service is designed to work without extra runtime dependencies. Supported devices include Android 6.0 and later, in both rooted and non-rooted modes.

Remote access can use the browser-based WebUI and supports MJPEG, H.264, and WebRTC streaming. WebRTC deployments can configure STUN and TURN servers. Remote desktop and RPC connections support end-to-end TLS and service-certificate access control, while the WebUI can use a custom login password.

Network configuration includes HTTP, HTTPS, SOCKS5, and Shadowsocks proxies, OpenVPN, frp forwarding, DNS and UDP proxying, LAN bypass, and automatic proxy connection at startup. mDNS can expose devices through `{device_id}.local` on a LAN.

## Tools and capabilities

- Automate controls with selectors, waits, swipes, scrolling, Unicode input, and coordinate queries.
- React to UI changes with watchers that trigger clicks, key events, or counters.
- Operate isolated virtual displays and switch among displays in the WebUI.
- Capture traffic with one-click MITM setup, per-package filtering, request and response editing, and upstream proxies.
- Use bundled Frida support, persistent scripts, hot reload, and Python or HTTP JSON-RPC calls.
- Inspect and control devices through a browser remote desktop with file transfer, terminal access, clipboard sharing, and live audio on Android 10 or newer.
- Use built-in ADB, SSH, SCP, proxy, VPN, and distributed networking functions.

## Limitations and notes

Some capabilities depend on the Android version: live audio requires Android 10 or newer, and coexistence with other accessibility services is supported from Android 8.0 onward. OCR model placement and GPU use can be centralized in clustered deployments, while image matching is performed on the device. WebRTC connectivity across networks may require suitable STUN or TURN configuration. Applications with custom-rendered interfaces may need OCR or image matching instead of standard selectors.

_Full upstream README: https://allmcps.com/mcp/lamda/readme_

