Agent loop
Plan first, use checkpoints, redirect Agent while it works, choose a Run Mode, and run Agent Review before you commit.
-
Which Run Mode Should You Use in Cursor? Run Modes decide when Agent can run shell commands, MCP tools and fetches without asking you. Level 3: Workflows
- 1Auto-reviewallowlist, then sandbox, then classifier
- 2Allowlistyour list runs, the rest asks
- 3Run Everythingno prompts, no sandbox
-
How Do You Keep Cursor's Agent on Track With a Plan and Checkpoints? Press Shift+Tab in the chat input until the mode shows Plan, describe the task, answer its questions, edit the plan, and click Build. Level 3: Workflows
- 1Clarify and research
- 2Plan
- 3You edit it
- 4Build
- 5Revert, refine, rerun
Then repeat -
How Do You Redirect Cursor's Agent While It's Working? You don't have to wait for Agent to finish. Level 2: ContextAgent keeps goingAgent stops
- Queue (Enter)
- Send now (Cmd+Enter)
- Stop
-
How Do You Get a Code Review From Cursor Before You Commit? Agent Review is a dedicated code review on your local changes. Level 3: WorkflowsAgent Reviewbefore you commit
- Reviews local changes
- Quick for small diffs
- Deep for sensitive code
Bugboton the pull request- Reviews on push
- Starts at Premium
- Ignores .cursor/rules
Three points where you can step in
You can shape a run before, during and after it. Before: Plan mode and your Run Mode. During: queue a message, steer it, or click Stop. After: read the diff, restore a checkpoint or run Agent Review. Most lost time comes from only stepping in at the end, when a wrong assumption has already spread.
Give it a finish line
An agent needs a way to know it's done. A command that passes is better than a description, such as "npm test passes and the list excludes deleted orders". Cursor's course says if the agent can't verify its output, you end up spending more time making corrections. For a long objective that spans many turns, use /goal.
How do I use Plan mode in Cursor?
Open Agent, press Shift+Tab until the mode shows Plan, and describe your task. Agent researches and asks questions, then produces a plan you can edit. Click Build when you're happy with it.
Can I send a message while Cursor's agent is working?
Yes. Enter queues it for after the current task. Cmd+Enter or Ctrl+Enter sends it immediately. Steering with Send now delivers it at the agent's next tool call where available.
What is a checkpoint in Cursor?
It's a local snapshot of modified files that Agent makes automatically before significant changes. Restoring one reverts files but keeps the conversation. Checkpoints are separate from Git, so commit anything you want to keep.
Where are Cursor plans saved?
In your home directory by default. Click Save to workspace to move a plan into your project for future reference and team sharing.