github.devotd.
NewRobotics

knifelemon/droidline — Android Automation Controlled from Code Without ADB

The public status of knifelemon/droidline, a Go tool that drives real Android devices from code without ADB or a USB cable, and its latest release v0.1.11.

Repositoryknifelemon/droidline ↗

The public status of knifelemon/droidline, a Go tool that drives real Android devices from code without ADB or a USB cable, and its latest release v0.1.11.

GitHub description: Android automation from code. Test your apps on real phones, script the tasks you repeat, run a shelf of phones, or let an AI agent use one, from Python, Node.js, the CLI, HTTP or Appium clients. No ADB, no USB cable, no root.

stars
★ 308
Forks
27
Language
Go
License
MIT
Latest release
v0.1.11
Read original source ↗View on GitHub ↗

knifelemon/droidline is a Go project that operates real Android devices from code. It supports app testing, scripts for repetitive work, controlling multiple devices, and device operation by AI agents. It needs no ADB, no USB cable, and no rooting, and commands can be sent through any path: Python, Node.js, CLI, HTTP, or an Appium client. The latest public release is v0.1.11, and its contents and the basis for its features are documented in the GitHub repository's README and release notes.

Command definitions are created from a single spec and generated in bulk into Python, Node.js, CLI, and HTTP bindings, so names and behavior are the same whichever path is used. On the device side it implements the accessibility service and the input method directly, and the connection layer adds end-to-end encryption and a reconnection cache. Existing Appium is tied to developer options and ADB, which made it hard to reach device farms or AI agents; this project removes that constraint. The license is MIT, the implementation language is Go, and the repository has 304 stars.

That said, the release notes do not make unfounded claims about measured device figures. Real processing speed and stability have to be verified separately. The repository contains unit tests that validate cipher vectors, accessibility matchers, and command spec drift, along with its own benchmark scripts and measurement result JSON, so there is room to check the performance claims again. The research record scores freshness 7, technique 7, quality 7, and momentum 5.8, and since the project was made public in October 2026, for now it stays close to the range of things worth watching lightly.