There are several voice browser implementations for dialog systems, but none of them are both open-source and standards-compliant, while retaining compatibility with multiple implementations of system components such as natural language understanding (NLU) and dialog management modules. We present an standards-compliant open source solution that closes this gap while incorporating support for modern dialog concepts like flexible switching of user goals, custom grammar design and adaptivity to users. We show that our implementation can flexibly interface with two different NLU implementations to extract semantic information from user input and expose it to a VoiceXML application which integrates into a cloud-based dialog system that handles real user traffic.