Question

A developer at a fintech company is working on a large Python file that references helper functions defined in other files they have open in adjacent editor tabs. They notice that when those related tabs are open, Copilot's inline suggestions become noticeably more accurate and consistent with their internal naming conventions. A teammate asks how Copilot achieves this without the code ever being manually pasted into the current file. Which explanation best describes what is happening during prompt construction?

Accepted Answer

Copilot uses a technique like neighboring/open-tab context to gather relevant snippets from other open files and includes them in the prompt sent to the model.

Answer

Copilot fine-tunes its underlying model on the developer's open files in real time, permanently improving the model for that repository.

Answer

Copilot indexes the entire repository on GitHub's servers and retrieves matching code from that index for every keystroke.

Answer

Copilot stores the contents of all open tabs in the completions log so future suggestions can reference them across sessions.