// the find
microsoft/HydraLab
Intelligent cloud testing made easy.
HydraLab is a self-hosted mobile and cross-platform test farm from Microsoft — you bring the devices, it manages them and runs Espresso, Appium, XCTest, or Maestro tests through a web UI. It's for teams that have physical test devices sitting around and want their CI pipeline to actually use them rather than paying per-minute for cloud device rental.
The center-agent architecture is clean: a central Spring Boot server coordinates multiple agent nodes, each of which manages its own attached devices — you can scale horizontally by plugging in more machines with more phones. The Uber Docker image is a genuinely nice touch for evaluation; one `docker run` and you have a working instance. Framework coverage is solid — Espresso, Appium (Java), XCTest, Maestro, and Python runners means it won't immediately reject your existing test suite. The APK static analysis tooling (apk-canary, apk-leaks) bundled into the agent is something most test farms don't bother with.
Spring Boot 2.2.5 is listed in the badge and Appium 8.0 is years behind current (Appium 2.x has a completely different plugin model) — unclear how much of this has actually been updated vs. left stale. Community size is thin for a Microsoft repo: 1,026 stars and 106 forks after what looks like years of development suggests it hasn't caught on outside internal Microsoft teams, so expect to debug things yourself. The 'TestGPT' smart explorer feature is mentioned prominently but the actual LLM integration looks like a thin wrapper around ADB monkey with some GPT-generated action sequences — don't expect the AI claims to hold up under scrutiny. Setup for iOS is genuinely painful: it requires macOS agent nodes with physical device provisioning, and the docs lean on 'see the wiki' for the hard parts.