Use official SDK as low-level implementation, or clarify advantages of not doing so

I think it makes sense to either use the official AWS SDK as the actual implementation of talking to AWS, or clarify the advantages of avoiding that implementation. It would also be helpful to understand whether the goal is to pursue new features, or just maintain knox for existing apps on the assumption that most users are migrating to the SDK.

I'm suggesting these questions be addressed in the documentation.

Review the existing Knox documentation and its AWS implementation, then compare them with the official AWS SDK. Document the advantages and trade-offs of the current approach, and clarify whether the project is pursuing new features or primarily maintaining existing applications.

Written by the indexing model from the issue text.