Abstract: According to the interactive requirements of rich media scene, a script for control of rich media scene in embedded environment. In this script, XML markup language is adopted to describe grammar. The semantics is represented by defining elements and attributes. The script can be executed by traversing the scene DOM tree and it resolves logic controlling, state recovering, and terminal device capabilities calling of the scene. The script can share lots of same modules with the scene, such as parsing modules and executing modules. Therefore, the script has the characteristics of high execution efficiency and low consumption of resources compared with some other existing scripts.