I’d like to request an option in Codex settings to make Ctrl+Enter always send and Enter always insert a newline.
Right now, the send behavior changes depending on whether the input is single-line or multi-line:
when the message is a single line, pressing Enter sends immediately
when the message becomes multi-line, the shortcut changes and Ctrl+Enter sends instead
This inconsistency is easy to trigger by accident, and I often send messages before I intend to.
My typical workflow is writing detailed prompts and task descriptions, so most of my inputs are multi-line. Because of that, I strongly prefer a consistent editing model where:
Enter always creates a new line
Ctrl+Enter always sends
this behavior can be enabled as a setting and stays fixed regardless of whether the current text is one line or multiple lines
A configurable option like this would make drafting long prompts much safer and more predictable.
You are welcome to post feature request here however the official place to request codex enhancements is the GitHub OpenAI Codex issues tab and use the label enhancement
If the matches what you seek then give the issue a thumbs up to let the OpenAI developers know this is important to you. If this does not match your desired request then look further or open a new issue with the label enhancement.
does this meaningfully change anything - how much is the LLM going to be swayed by a NL char? ie can the prompt really improve from adding NLs?
I suppose it avoids having to cut and paste specifications from another text editor and you can then write them with headings directly inside the CLI? … just curious