Support multiple --workspace paths in Cursor Agent CLI

Ideas Feature Requests
, ,
1

Feature request for product/service

Cursor CLI

Describe the request

Currently, the Cursor Agent CLI only accepts a single --workspace path.
This forces developers who work across multiple repositories to use a
large parent workspace that loads hundreds of skills and rules into
context, even when the task only touches one specific repo.

The request is to allow multiple --workspace flags in a single agent call:

cursor --workspace ./repo-a --workspace ./repo-b …

This would allow the agent to have access to multiple repositories
simultaneously without inflating the context with irrelevant skills
and rules from unrelated projects.

Use case: in a monorepo-adjacent setup where a code generator (repo A)
produces output consumed by platform repos (repo B, C, D…), it is
common to need the agent to read from the generator and write to the
platform, but loading the entire parent workspace just to achieve this
wastes significant token context on every single invocation.

5

same request

6

Hey @Jefferson_Motta, thanks for the suggestion! Good news, you can do this today.

The CLI has a repeatable --add-dir flag that adds extra workspace roots next to the primary one. Rules and skills are loaded from each root you pass, not from the parent folder, so you only get the context for the repos you name:

agent --workspace ./repo-a --add-dir ./repo-b --add-dir ./repo-c

Once you’re in a session you can also run /save-workspace my-platform, and next time just launch agent --workspace my-platform to get the same set of repos back.