Expose reliable Cursor agent status events for integrations

Feature request for product/service

Cursor IDE

Describe the request

I’m trying to build some small external integrations around Cursor agents, both in the IDE and potentially through Cursor CLI/headless workflows.

Right now, the current ways to detect agent state don’t seem 100% reliable, especially if you want the same logic to work across the IDE and terminal/CLI usage. You often have to infer state indirectly from UI changes, timing, terminal output, process state, or other heuristics.

There are quite a few useful things this could enable: LED strip status indicators, small desktop widgets, physical buttons/devices, notifications, or other helper tools that react to what the agent is doing.

A simple, stable and documented agent status event would solve a lot of this.

Something like:
working — agent started or resumed
waiting — needs user input, approval, or AskQuestion
done — turn finished and agent is truly idle
error / cancelled — optional

Ideally this would be available for both the Cursor IDE and Cursor CLI/headless mode, so external integrations can react reliably without depending on brittle UI or terminal-output detection.

(translated from Dutch to English by ChatGPT, for better readability)

Operating System (if it applies)

Windows 10/11
MacOS
Linux

Hey @Agenthsmith!

Thanks for the feedback. Have you had a chance to look at Hooks? They work across both the IDE and CLI, and signal when something has happened.

It’s not quite as simple as a single status, but they are deterministic.