Title:
Add opt-in “auto-accept” GPT suggestions feature via JSON settings (not UI)
Category:
Feature Requests
Body:
Hi Cursor team,
We understand that auto-applying GPT suggestions without user confirmation is intentionally disabled to protect against accidental code modifications -and that’s a perfectly reasonable default.
However, for advanced users or fast workflows (like code generation from structured prompts, e.g. Swagger → Controller/DTO), it would be extremely helpful to have an opt-in auto-accept mode. This mode could be hidden behind a settings.json flag- for example:
"cursor.experimental.autoAcceptAllSuggestions": true
This way, users would need to manually modify the JSON settings file, acknowledging the risks and making it clear that they take full responsibility for the results.
It’s okay if such a mode skips safety checks, UI dialogs, or is entirely undocumented- as long as it works deterministically and doesn’t require mouse interaction for fast batch operations.
We’re not asking to enable it by default or expose it in the UI -just to allow it for those who know what they’re doing.
Thanks for building such an amazing tool 
Happy to beta-test any version if needed!