The term procedural content generation (PCG) refers to writing software which can synthesise content for a game (or other media such as film) without further intervention from a designer. PCG has become a rich area of research in recent years, finding new ways to apply artificial intelligence to generate high-quality game content such as levels, weapons or puzzles for games. Such research is generally constrained to a single type of content, however, with the assumption that the remainder of the game's design will be fixed by an external designer. Generating many aspects of a game's design simultaneously, perhaps ultimately generating the entirety of a game's design, using PCG is not a well-explored idea. The notion of automated game design is not well-established, and is not seen as a task distinct from simply performing lots of PCG tasks at the same time. In particular, the high-level design tasks guiding the creative direction of a game are all but completely absent in PCG literature, because it is rare that a designer wishes to hand over such responsibility to a PCG system. We present here ANGELINA, an automated game designer that has developed games using a multi-faceted approach to content generation underpinned by a co-operative co-evolutionary approach which breaks down a game design into several distinct tasks, each of which controlled by an evolutionary subsystem within ANGELINA. We will show that this approach works well to automate game design, can be ported across many game engines and game genres, and can be enhanced and extended using novel computational creativity techniques to give the system a heightened sense of autonomy and independence.