It’s being touted as a killer feature, but I find the inline “explain code” pop-up in Windsurf more disruptive than helpful. It triggers automatically when you hover over certain code blocks, and the explanation is often just a rephrasing of the syntax you’re already looking at. It feels like having an overeager intern constantly leaning over your shoulder to state the obvious.
Maybe it’s useful for absolute beginners, but for anyone doing serious work, it breaks flow. I’m trying to audit a chunk of authentication logic, and a tooltip keeps popping up to tell me “this is a function that returns a token.” No kidding. I’d rather that screen real estate and processing power went toward something actionable, like highlighting potential insecure patterns or flagging deviations from our internal standards.
Worse, I can’t find a way to disable it permanently without turning off all hover-based features. In a security context, you want clarity and control, not AI-powered guesswork about what you might not understand. It starts to feel like a gimmick that’s more about marketing the “AI” than actually improving the review process.
Am I the only one who just wants to read the code without commentary? Or is there a hidden config to kill this thing for good?
Trust but verify
No, you're not alone. I've seen this pattern play out across several clients who've trialed similar "intelligent" IDE features. The core issue is that these features are designed for discoverability, not for sustained, focused work.
You've hit on the critical point about security auditing. When you're reviewing authentication logic, your mental model is about data flow and edge cases, not syntax definitions. An unwanted popup explaining a function signature is cognitive noise that can actually cause you to miss a subtle logic flaw. It's the equivalent of someone tapping your shoulder while you're checking a lock.
The workaround I've suggested, and it's only a workaround, is to create a custom editor profile with all "assistance" triggers disabled, and switch to it for review sessions. It's a nuisance, but it's better than nothing. The real fix is for vendors to add a single granular toggle for "auto-explain," but that doesn't seem to be a priority. They're selling the "wow" factor, not the "don't interrupt me" factor.
Mike
Agreed, especially about the security review context. When I'm mapping out OAuth flows for an integration, my focus is on the data path and potential leak points. A popup explaining what a variable is adds zero value and can actually mask the line I need to see.
You mentioned wanting it to flag deviations from internal standards instead. That's the real need. A tool that could cross-reference a code snippet against a security ruleset you define, like flagging a hardcoded secret in a PR, would be genuinely actionable. The current "explain" feature feels like a solution looking for a problem.
I had to dig into a config file to turn it off. Look for "inlineHints" or "codeLens" in the settings JSON. Disabling that specific provider was the only way.