# Linking Workspaces

Linking lets you give an agent access to another workspace's context. This is super handy when you need to coordinate work across repositories or have one agent aware of what another is doing.

## How to Link

You can link workspaces by pressing on the **+** icon in the chat window and selec the workspace. It usually makes sense to ceate a fresh workspace if you expect to also make changes in this new workspace or to just give it a fresh context.

![Linking a workspace with @mention](/images/docs/link-workspaces.png)

## Use Cases

### Cross-Repository Changes

If you're building an API in one repository and a frontend that consumes it in another, link the workspaces so the frontend agent can see the API's code and types. This avoids mismatched contracts and duplicate effort.

### Coordinating Changes

If two workspaces are making related changes (e.g., one updates an API and another updates the client that calls it), linking lets the agents stay in sync without conflicting.

### Supervisor Pattern

Create a "supervisor" workspace that reviews or orchestrates work happening in other workspaces. Link the supervisor to the worker workspaces so it can monitor progress and provide guidance.

### Cross-Referencing

If one workspace has already solved a similar problem, link to it so the agent can reference the approach without duplicating effort.

## Things to Know

- Linked workspaces can **read** each other's state but work in their own isolated clones
- The link is **per-prompt** — you choose which workspaces to reference each time
- Linking works across repositories, not just within the same repo
- This is most useful for coordination and context sharing, not for sharing file changes between workspaces