Entity Framework Core 1.0 supports two different workflows:
The first approach was made popular when Entity Framework 4.1 "Code First" was released. The idea here is that we, as C# developers, understand code better than anything, and so we model our entities as code. It is left for Entity Framework to produce the database objects (tables, relations, and so on) that will enable it to persist our data in an almost transparent way.
But what happens when you have an existing (legacy or otherwise) database, perhaps consisting of hundreds of tables? Or you have this super cool UML tool that generates the database from a model? It is troublesome, to say the least, to generate all C# entities by hand in Visual Studio, so the option is...