AI Catchup

Cursor Cloud Agents Add Event Triggers, Long-Lived Goals, and Isolated Subagents

By 4 min read

Cursor says Cloud Agents can now pick up work from events, keep working toward a long-lived goal, monitor pull requests, watch Slack threads, run scheduled tasks, and launch subagents in isolated virtual machines. The same update adds skill-based Custom Modes and less disruptive steering while an agent is working.

Cursor's latest Cloud Agents update moves the product toward event-driven, persistent workflows. In an August 19 announcement, Cursor said agents can pick up work in response to events, hold a goal until it is met, and stay on course through long-running sessions. (Cursor on X; Cursor changelog)

The update is broader than one new command. Cloud Agents can now monitor pull requests, watch a Slack thread, or run scheduled tasks; Cursor also added isolated virtual machines for subagents, skill-based Custom Modes, and steering that waits for the next tool call instead of interrupting an action. (Cursor on X; Cursor changelog)

Event-driven Cloud Agents

Cursor says a Cloud Agent can subscribe to an event source such as a thread or conversation and wake when something happens. The examples in the changelog include:

  • Monitoring a pull request.
  • Watching a Slack thread.
  • Running a scheduled task.

Cloud Agents automatically subscribe to pull requests they create and can drive those changes toward completion by fixing CI failures and addressing bot comments, according to Cursor's changelog. Subscriptions are available for cloud agents only for now. (Cursor changelog)

This changes the workflow from starting an agent for every discrete request to giving it a place to keep watching. A scheduled task or a follow-up comment can become the event that resumes the work, while the changelog's cloud-only qualification is an important boundary for teams comparing local and hosted agents.

Long-lived goals with /goal

The new /goal command gives an agent a long-lived objective to work toward until it is fully complete. Cursor's example is /goal fix all flaky tests and make CI green; the changelog says to use it in a new chat. (Cursor on X; Cursor changelog)

Cursor says /goal can be paired with a Custom Mode for following a playbook and with /loop for recurring check-ins. That combination is aimed at work that needs continuity rather than a single prompt-and-response exchange. The announcement does not promise that every goal will finish without review, so treat the command as persistent task direction, not an autonomy guarantee.

Subagents get isolated machines

Subagents can now run on their own virtual machines. Cursor says each one receives an isolated copy of the project, clean context, and its own cloud environment. (Cursor on X; Cursor changelog)

The intended uses are parallel and adversarial: have subagents test the parent agent's changes in fresh environments, or swarm independent fixes without collisions. Separate machines make that workflow easier to reason about because a test agent does not need to share the parent agent's working state.

Custom Modes and less disruptive steering

Any skill can now be used as a Custom Mode, which Cursor describes as a skill that stays pinned in the chat. From the / menu, choose a skill and press ⌥⏎ on Mac, Alt+Enter on Windows, or select Use as Mode. (Cursor on X; Cursor changelog)

The same release changes how follow-up messages steer an agent. Steering waits for the next tool call instead of cutting the agent off mid-action; Cursor says users can type a follow-up and press Send, or press Enter twice. (Cursor on X; Cursor changelog)

What this means for a team

The pieces fit together as a more persistent agent loop:

  1. Give an agent a goal.
  2. Pin the operating instructions as a Custom Mode.
  3. Let an event, schedule, or pull-request update wake the Cloud Agent.
  4. Use isolated subagents for testing or parallel work.
  5. Steer the run without stopping the current tool action.

That is a meaningful workflow improvement for teams already using Cloud Agents, but the official changelog does not establish a new pricing tier, universal availability, or local-agent parity. The clearest published limitation is that subscriptions are cloud-only for now. (Cursor changelog)

Sources

Keep building the workspace playbook

Frequently Asked Questions

What can Cursor Cloud Agents respond to now?

Cursor says Cloud Agents can monitor pull requests, watch a Slack thread, and run scheduled tasks. Agents can subscribe to an event source and wake when something happens.

What does /goal do in Cursor?

The /goal command gives an agent a long-lived objective to work toward until it is complete. Cursor shows it being used for a task such as fixing flaky tests and making CI green.

How do Cursor subagents run in isolation?

Cursor says each subagent can run on its own virtual machine with an isolated copy of the project, clean context, and its own cloud environment. That lets subagents test a parent agent's changes or work on independent fixes without collisions.

Are Cloud Agent subscriptions available everywhere?

Cursor's changelog says subscriptions are available for cloud agents only for now. The announcement does not state a plan change or pricing change for these features.

Get the weekly AI Catchup

Tools, practices, and what matters, in your inbox every week.