implementation("io.github.brianmwas.koog_compose:koog-compose-mediapipe:2.1.0")
On-device AI provider for koog-compose using Google's LiteRT-LM engine. Runs quantized Gemma models directly on Android with no network required.
