ghc-chat-driven-debugging

Highlight the broken code, run /explain to confirm Copilot understands it, then use /fix with a specific steering instruction rather than a bare command. Review the proposed diff before accepting, and re-run the program to verify the fix against the original failure and new edge cases.

/explain describes what the selected code does and often flags risks, while /fix proposes an actual code change as a reviewable diff. Running /explain first verifies Copilot understood the code the way you do, since a wrong explanation leads to a wrong fix.

Copilot's first instinct is often to make the error go away, which suppresses the symptom rather than fixing the cause, such as silently dropping a negative deposit. Push back in chat with what is wrong with the fix, for example asking it to raise a ValueError so the failure stays visible.

Yes, but expect a multi-turn conversation. Ask /fix to make the transfer atomic, then follow up with questions like what happens if the rollback itself fails, which usually steers Copilot toward upfront validation before any mutation, the cleanest pattern.

You need to have completed the ghc-copilot-foundations project so the Chat panel and slash commands work, plus Python 3.10 or later and a terminal such as pwsh or the built-in VS Code terminal to run the buggy program.

Use the VS Code debugger with breakpoints and stepping when you need to inspect runtime state line by line; it complements the Copilot Chat loop rather than replacing it. This project deliberately excludes debugger training and test writing, which are covered in separate projects.