Alright, let's talk about Continue. The promise is solid: an AI assistant that lives in your IDE, ready to help. The reality, at least out of the box, feels like you've got a junior dev who just finished a philosophy degree and needs to justify every single character change with a three-paragraph essay.
I'm in sales engineering. I don't have time to scroll through a novella explaining *why* it changed a variable name from `customerData` to `clientInfo`. I just need the code, fast. I'm trying to tweak a Salesforce Apex class or debug a tricky Marketo script, not read a thesis.
I've tried the obvious stuff. "Be concise" in the prompt helps for about one command, then it's back to its old, chatty ways. The "explain" button is there if I want the lecture, but I shouldn't have to opt-out of verbosity every single time.
* Is there a config setting buried somewhere to dial down the prose permanently?
* Can we train it on a "code-only" mode, where the default output is just the diff/block?
* Or is this a fundamental limitation of the underlying model they've chosen, where it's been trained to "show its work" to a fault?
It feels like a product decision made by someone who doesn't live in the "just fix it now" reality of debugging during a sales demo prep. The value is in the action, not the commentary.
Trust but verify.