Providing secure input and output to a trusted agent in a system with a high-assurance execution environment

Techniques are disclosed to provide security for user output and input in which a first, host operating system is used along with a second, high assurance operating system (nexus), where the first system provides at least some of the infrastructure for the second system. A trusted UI engine has a trusted input manager and a trusted output manager. The trusted input manager controls access to trusted input, distributing decrypted input to the host operating system where appropriate, or to the appropriate process running in the nexus. The trusted output manager manages output to the display, and allows trusted agents in the nexus to output data for display without needing to be aware of output-device-dependent details.