Latest in AI
Kiro IDE Remote Workspaces: Dev Container, SSH or Open in IDE?
Choose where your code, terminal and agent run in Kiro 1.5.11, and verify the workspace before assuming it matches your local checkout.
Kiro IDE 1.5.11 added Remote SSH, Dev Containers and the ability to open an existing Kiro Cloud Session in the desktop editor on November 20. All three can give you a familiar editing window. They put the terminal and agent in different places, which matters when you need the right dependencies or want to continue a specific conversation.
Start by deciding which environment has the state you want to keep. Opening a cloud session does not bring it into a local checkout, and reopening a repository in a Dev Container does not attach it to a cloud conversation.
Pick where the work should happen
| What you need | What to open | Where the project work runs |
|---|---|---|
| Use tools already installed on a Linux machine you manage | Remote SSH | On that Linux host |
| Use the project’s declared dependencies | Dev Containers: Reopen in Container | In a container on your computer or your SSH host |
| Continue a Kiro Cloud Session already working on the project | Open in IDE | In the existing Kiro-managed cloud sandbox |
If a teammate's test passed only in a specific container image, start with the repository's Dev Container. If the pending changes only exist in a cloud session, open that session instead. Picking the Git branch in a new local window will not recover changes that have not been committed or pushed.
Reopen a trusted project in a Dev Container
Kiro's Dev Container guide documents two accepted config locations: .devcontainer/devcontainer.json and .devcontainer.json. A local container needs a running Docker-compatible engine. If you connect to a remote Linux host first, Docker runs on that host and must be available on its path. Installing Docker on your laptop does not fix a missing remote Docker command.
Review the configuration before choosing Dev Containers: Reopen in Container. In particular, an initializeCommand runs on the local or SSH host, outside the container. A container does not make setup from an unfamiliar repository safe. Start with a project and host you control, and use Dev Containers: Show Container Log to distinguish a build failure from an IDE or sign-in issue.
Inside the container window, check the working directory and the language runtime's version, and run a small existing test. Kiro gives the container window its own sign-in and chat history. That explains why a conversation from the local or plain SSH window might not be there.
Open an existing cloud session without cloning locally
In Kiro IDE, open Agent Focus, find the session under Cloud Sessions, and choose Open in IDE from its actions. The Cloud Sessions guide says you need an eligible plan, the signed-in account and access to the session. You do not need an SSH key or local container engine for this route.
In the new window, check the current branch, working directory and any outstanding edits before continuing. Closing the window does not stop the cloud session. Reopening it should reconnect you to the same session rather than creating another one.
The editor window is on your computer, but the cloud workspace is not. Kiro's guide says prompts, attachments and relevant code context go to its services. If your team has not approved that placement for the repository, use an approved workflow instead and confirm its data handling separately. Where commands execute and where model requests go are two different questions.
For the first trial, I would open a small test project and verify the branch, runtime version and conversation in one window. Most handoff confusion starts when one of those changes unnoticed.
I'm Paras. I help AI teams explain new developer workflows in practical technical writing.