Cursor CLI Session Management and Cloud Execution

05-tips already mentioned cursor agent --resume to continue the last conversation. This post completes sessions and cloud execution: how to list past conversations, which one to resume, and using -c to run a task in the cloud so you can close the terminal or laptop and it still finishes in the background.

Session resume: --resume

You closed the terminal mid-conversation, or the task was too long and you stopped it—next time you want to continue from the same thread:

cursor agent --resume

This loads the most recent conversation; context and previous instructions are still there, no need to re-describe the request.

Resuming a specific conversation

If you have more than one conversation and want to resume a particular one:

cursor agent --resume=chat-id-here

Get the chat-id from “list conversations,” below.

List all conversations: cursor agent ls

cursor agent ls

This lists past conversations (with ID or identifier). Find the one you want and pass its ID to --resume=ID.

Quick reference:

Command	Purpose
`cursor agent ls`	List all conversations
`cursor agent --resume`	Resume the most recent
`cursor agent --resume=chat-id`	Resume a specific conversation

Cloud execution: -c / --cloud

Some tasks are heavy: full-repo analysis, large doc generation, multi-step refactors. Running locally means keeping the terminal open; closing the laptop or losing connection stops the run.

With -c (or --cloud) you hand this conversation to the cloud Agent:

cursor agent "Analyze the whole project and write architecture doc to docs/architecture.md" -c

After you send it, the task runs on Cursor’s cloud
You can close the terminal, shut down, or switch to something else
Later, from any device with the same Cursor account, use --resume to rejoin that conversation and see the result or add more instructions

So -c = “run this in the cloud; I don’t have to wait at the computer.”

When to use -c?

Situation	Suggestion
Task will take a long time	Use `-c` so it runs in the cloud; it can finish after you close
Quick one-off question or small edit	Skip `-c`; run locally
Unstable network or prefer not to rely on cloud	Run locally, but keep the connection up

Sessions + cloud together

A typical flow:

Start a task with -c so it runs in the cloud
Close the terminal
Later run cursor agent ls to find that conversation
cursor agent --resume=that-id to rejoin, see the result, or send more instructions

That way you cover both “long-running tasks” and “multiple devices / shutdown.”

Summary

cursor agent ls: List all conversations so you can find the one to resume
--resume: Resume the latest; --resume=ID: Resume a specific conversation
-c / --cloud: Run the current task in the cloud; you can close the terminal and resume later with --resume

This wraps up the series. Review from 00-overview, or revisit 08-worktree-merge and 09-worktree-advanced to master Worktree and Best-of-N.