Hi everyone. I’m fairly new to Continue and still figuring out how to make it work best for my workflow. I’ve been using it for about a week while building some email campaign templates and setting up a simple CRM integration.
I really appreciate the powerful suggestions, but sometimes they’re a bit too much for what I need. For instance, when I’m just writing a straightforward Python script to clean a mailing list, Continue might suggest a whole refactored class structure when I just wanted a line to remove duplicates. It’s impressive, but it breaks my flow.
My question is: is there a way to configure Continue to give me simpler, more direct code completions most of the time? I’d like to filter out or tone down the larger, more architectural suggestions unless I explicitly ask for them. I’ve looked through the settings but couldn’t find a clear option for this.
I work mostly on landing pages and campaign analytics, so my code is often simple scripts for data handling or HTML snippets. Having a “basic mode” would be really helpful for my learning pace, as I like to understand each step fully before moving to complex patterns. Any guidance on settings or workflows that could achieve this?
You're hitting on the fundamental issue with these AI assistants, and it's not really a configuration problem. The model is trained to produce "complete" solutions, often interpreting a simple request as an invitation to over-engineer. It's the same impulse that turns a script into a microservice.
> I'd like to filter out or tone down the larger, more architectural suggestions
There's no slider for that because the complexity is baked into its worldview. You can sometimes get simpler completions by being brutally specific in your comments. Instead of just typing a function name, write a comment like "# remove duplicate emails from list, in place, simple loop". It might still give you a lecture on pandas, but it sometimes helps.
Honestly, for your use case with HTML snippets and data scripts, you might be better off using a traditional, dumb autocomplete that just finishes the line you're on. These tools are fantastic when you want the fancy suggestion, but they're awful at staying out of the way.
monoliths are not evil