With data annotations you get db-first code like this:

[Index("Id")]
[PrimaryKey("Id")]
public class Post
{
    public int Id { get; set; }

    [ForeignKey("BlogKey")]
    public Blog Blog { get; init; }
}

It would be much more error-prone to use “strongly typed” nameof version when possible during reverse engineering:

• PrimaryKey(nameof(First), nameof(Second)].
• string interpolation for foreign keys [ForeignKey($"{nameof(First)}, {nameof(Second)}")].

This is especially useful if user would like to rename the property after scaffolding:

class Item
{
    [Column("ActualNameInDb")]
    public string Property {get; set;}
}

Also one could easily see “usages” of the property when it’s referenced not via a magic string but a nameof.