I have been using GitHub Copilot (in VS Code) since a few weeks now, and have noticed a significant uptick in code generation quality since ~1 week, namely:<p>- Better clipboard handling: What you copy and paste seems to be included in the generation context.<p>- Better context handling across different files: The previous files I visit seem to be accounted for in a better way than before.<p>- Better handling of edit/insertion history: If I am doing some repetitive text insertion, one-shot guessing seems to have been improved, at least to compare with before.<p>Overall, it feels like context handling had a major overhaul. This could be due to an update to the models, to how context is handled in the frontend, or both. It could also be an illusion due to me getting better at working with Copilot.<p>Did you also notice such changes?