"Add to queue" chat requests, non-blocking

Feature Requests
1

As of (0.45.11) you can’t submit a request without interrupting the running one.

But, you should be able to queue multiple prompts, just by hitting Cmd+Enter or or pressing “Enqueue” and they would run sequentially.

This would save time for users that think/type faster than generations usually take.

6 Likes
2

+1, and my own elaboration:

Non-interrupting Command Queue for Cursor

Problem

When working with Cursor’s AI assistant, users often notice things they want to change or request while the AI is in the middle of executing a multi-step task. Currently, interrupting the AI mid-task can disrupt its flow and context. Users need a way to queue these requests without interrupting the current task.

Use Case Example

  1. AI is implementing a complex feature with multiple files

  2. User notices a minor change or task they want done - something that should be addressed but isn’t urgent enough to interrupt the AI’s current work

  3. User wants to queue this request without stopping the AI’s current work

  4. After current task completes, AI processes the queued requests

Proposed Solution

  • Add a command queue UI element (perhaps a sidebar or collapsible panel)

  • Hotkey to quickly add to queue (e.g., Cmd/Ctrl+Shift+Q)

  • Queue items should include:

  • Priority (optional)

  • Request text

  • Context when queued (file, line number, etc.)

  • AI processes queue when:

  • Current task completes

  • User explicitly requests queue processing

  • AI reaches a natural stopping point

Benefits

  1. Maintains AI’s focus on current task

  2. Preserves user’s thoughts/requests for later

  3. Improves workflow efficiency

  4. Reduces context switching

Implementation Notes

  • Queue should persist across sessions

  • Allow queue item reordering

  • Option to merge related requests

  • Clear visual indicator of queued items

1 Like
3

+1, this would be quite helpful, and I think about this every session I work with Cursor.

4

Cursor’s chat agent is a fantastic productivity boost, especially for longer-running tasks. However, currently, if you send a message while the Cursor agent is still executing, it immediately interrupts the current run.

To improve usability, it’d be helpful if Cursor provided a clear choice when sending messages during active agent runs. Specifically, clicking “Send” should prompt users with options like:

  • Interrupt current execution and send immediately
  • Queue message to send after current execution completes

I’ll often pre-write a follow-up message to the running Cursor agent and then wait to hit send on it until I hear the bell nootification that the agent has finished running. The messages are often along the lines of “Double-check your work” or “Ensure that your solution meets these specific acceptance criteria [LIST OF SPECIFIC THINGS FOR THIS TASK]”, etc. These follow-up messages are typically task-specific, and hence not generalizable to a cursor rule we can attach to all initial prompts.

1 Like
6

Often when editing files, I realize I want another additional feature implemented right after it has finished. I would love to be able to put these into a queue (kind of like “Send Later” functionality that many messaging apps has), such that the message gets sent to the agent right after it has finished its current task. That way I can better queue a lot of work to the agent, with less overhead of checking when it is done etc.

This requests seems to actually be a duplicate of Queue chat messages during active Cursor agent runs and "Add to queue" chat requests, non-blocking

1 Like
7

Merged 3 related feature requests into one, preserving the order and votes.

Add: Please note that multiple feature requests to same topic dilute the focus and attention. there are then too many posts to read for everyone :slight_smile: hope this helps.

2 Likes
8

This would be wonderful, I came here because I was trying to find if this was a thing. And I just signed up so I could +1 =)
Here’s another thought though, queueing would be great, but what would be even greater would be if you could also comment on the CURRENT running task while it’s still running. The agent stops for tool calls and to do some thinking all the time anyway, so why not have it check the chat at the same time to see if the user has any input it could make use of for the current task.

1 Like
9

Thanks for the great addition to the feature request :slight_smile:

It would be very practical to leave remarks for AI

1 Like
10

wow 7 days later and Eric shared a sneak peek :slight_smile:

2 Likes
11

This is now possible on version 1.4! Quick overview of the behavior:

  • The default behavior is Send , which will send the message at the next opportune time for the agent (to make sure we don’t interrupt it while doing things like editing a file).
  • You can use ⌥+Enter to Queue, which will tee the message up after the agent finishes its loop.
  • You can use ⌘+Enter to Stop & send, which will immediately stop the agent and send your message.
  • Your default behavior is set in Cursor Settings → Chat → Queue messages.

Let me know if you have any questions and hope this is useful for you :slightly_smiling_face:

2 Likes
12

@austinp I just tried queuing a message now (with ⌥+Enter) in 2.0 and it doesn’t seem to work. Is this a known change? If so, why? I can see cases where queuing is preferable over interrupting at the next opportune time. I will note that I am using the Background Agent in the Cloud, so maybe that has something to do with it