A media interaction system is described herein that allows hosted applications to consume and/or produce rich media content independent of the format of the media content, while still maintaining the security of the host. The media interaction system accepts raw, uncompressed media data from a hosted application, and allows the hosted application to provide its own codec through a sandboxed API that restricts the actions the codec can perform to a set of trusted actions. Then, the application provides the uncompressed data to the system for playback. Thus, the media interaction system provides rich media content in any format that the application developer supports, and maintains the security of the user's computer system.