Experimental C# AI that acts as an assistant to developers, coding alongside them and performing refactorings, refinements and code creation fully autonomously. It does this by taking source code and breaking it down in a way a traditional developer does - by trying to codify the 'mental summarizations' that many developers rely on day-to-day. These summarizations allow people to see phrases, data structures or concepts and then be able to reliably guess about what code is attempting to do, how it's built, and what sort of constraints it labours under.

For a precise example: developers know what a map/dictionary is, what it's good at, and how to use it, but trying to work these out by using static analysis is tricky, computationally lengthy, or just impossible. This project attempts to bridge this gap allowing for systems to understand source code in a way similar to a developer. Not just syntactically and semantically, but also architecturally and conceptually.

This will be highly experimental for a long time, but many of the algorithms and concepts will be written in documentation in the future, so please check back later if you're interested.

Actual work done in the develop branch.