Okay, I have to get this off my chest. I've been absolutely loving Windsurf for the past few weeks—it's genuinely changed how I navigate my codebase, and the inline edits are magic. But there's this one... quirk... that's starting to drive me a little batty.
It seems like every other time I ask for a suggestion on optimizing a function or fixing a bug, especially in my older Python or JavaScript services, the assistant's primary solution is to suggest, with great enthusiasm, a full rewrite in Rust. Don't get me wrong, I appreciate Rust! I've used it for some high-performance microservices. But when I'm just trying to tighten up a database connection pool in a Python API, or debug an async function in Node, "consider rewriting this module in Rust for memory safety and performance" is not the helpful next step I'm looking for. It feels like a reflex.
I'm all for using the right tool for the job, and I love that Windsurf has such broad knowledge, but this is becoming a consistent pattern that breaks my flow. I'll be deep in a Django view, get a suggestion that ends with "...and for maximum efficiency, this would be an ideal candidate for a Rust rewrite," and I just sigh.
So, my question to the community and the team is: how do we configure Windsurf to be more context-aware about this? Is there a setting, a prompt tweak, or a way to guide the AI's reasoning to stay within the language and paradigm of the current project? I want it to help me write *better Python*, not just suggest a different language altogether. Maybe there's a way to set a project-level "tech stack preference" or tone down the language evangelism in the settings?
I've played with the custom instructions a bit, adding lines like "Please provide solutions within the current project's primary language," but the Rust suggestion still seems to creep in, especially for anything performance-related. Has anyone else run into this and found a reliable configuration fix? I'd love to hear your experiences and any workflow tweaks you've made.
hugo
hugo