The present work is an attempt of designing an expert system shell for general diagnosing systems that contains all the facilities to help its users (human experts) to complete their jobs (build their own expert systems) easily and efficiently. Two shells were designed and implemented. The first is called GADES, which stands for General Automatic Diagnosing Expert System. While the second is called GACES, which stands for General Automatic Code Expert System. Both designed shells were constructed as two phases shell. These two phases are: automatic friendly interface (expert interface), and inference engine. The first phase is responsible for acquiring human expert knowledge in an automatic way by using menus and message windows. The second phase is concerned with the design of inference engine together with a user interface that uses a backward chaining method. The two designed shells consider reasoning under uncertainty using certainty factor method (Stanford Certainty Factor). The difference between the two designed shells is: GADES uses the external database approach to store the human expert knowledge in order to be manipulated by the inference engine, while the GACES store the human expert knowledge in a source code file (represents the inference engine) that can be executed in turbo prolog language. Further, to support understanding of the reached conclusions when using the designed expert system, GADES and GACES supports two types of explanations Why and How. Why concerned with why the system asks a specific question, while How concerned with how the system arrives at specific conclusion. GADES and GACES were implemented by using Turbo Prolog Version 2.0 for building the inference engine and Visual Basic Version 6.0 with its supplement tools such as help maker, which is used for building the expert interface tools.