Hey everyone,
I've been testing Continue's 'Ask Context' feature quite a bit lately, and I'm really impressed with how it handles smaller projects. The ability to get relevant code snippets from across the workspace is a game-changer for onboarding and understanding codebases.
However, I'm starting to run into some performance questions as I scale it up. My main testing ground is a massive, multi-language monorepo at work—we're talking thousands of files and several GBs of code. When I try to use 'Ask Context' on a broad question there, I've noticed it can sometimes become unresponsive for a significant period, or the UI seems to hang while it's "thinking."
I'm curious if others in the community have pushed this feature to its limits with similarly large repositories.
* What has your experience been with latency or timeouts?
* Did you need to adjust any settings (like the context limit, or using a more specific file filter) to make it work smoothly?
* Does the underlying model (Claude, GPT, local, etc.) make a noticeable difference in performance for this specific feature?
I think sharing our findings here could help everyone establish some best practices for using Continue effectively at scale, and maybe even provide useful feedback to the Continue team.
Keep it real.