How to use JetBrains Gateway with a cloud sandbox

Run runtime sandbox ssh config --install, then add a Gateway SSH connection to <id>.runtime, user runtime, using the OpenSSH config.

On Runtime the IDE backend runs in a microVM of its own, reached through Runtime's authenticated API, so no SSH port is open to the internet. You size the machine to the project, pay for the CPU it uses at $0.025 per active vCPU-hour plus $0.0075 per reserved GiB-hour of memory, and let it pause when you stop working (pricing). There is no plan fee: prepaid credit starts at $10, and new accounts get 50 free sandbox hours.

Connect

Once, on your machine:

Terminalnpm install --global withruntimeruntime sandbox ssh config --install    # every sandbox becomes <id>.runtime to OpenSSH

Then in JetBrains Gateway:

1. Under SSH, choose New Connection.
2. Host: <id>.runtime or <name>.runtime. User: runtime.
3. Authentication: Parse config file ~/.ssh/config, so Gateway uses the entry the CLI installed and its ProxyCommand.
4. Pick the IDE and the project directory, such as /workspace/app.

Gateway downloads the IDE backend into the sandbox and opens the project in JetBrains Client on your machine, which is the thin window you type in (JetBrains docs, read 25 September 2026). The backend lands in ~/.cache/JetBrains/RemoteDev/dist by default, which in a sandbox is under /workspace.

Size the sandbox for an IDE

An IDE backend indexes the whole project, and JetBrains recommends more CPUs and memory for larger projects (prerequisites, read 25 September 2026). Ask for it at create:

TypeScriptimport { Sandbox } from "withruntime";const ide = await Sandbox.getOrCreate("ide", {  vcpu: 4,  memoryMiB: 8192,  diskMiB: 20_480,  idlePauseSeconds: 1800,});console.log(`${ide.info.name}.runtime`);

Pythonfrom withruntime import Sandboxide = Sandbox.get_or_create("ide", vcpu=4, memory_mib=8192, disk_mib=20_480,                            idle_pause_seconds=1800)print(f"{ide.info['name']}.runtime")

Terminalruntime sandbox create --name ide --get-or-create --vcpu 4 --memory 8192 --disk 20480 --idle-pause 1800

Indexing is the busy part; once it is done, a backend that waits for you uses little CPU, and CPU is billed on what is measured.

Keep the project warm

A named sandbox with an idle pause is the same workspace every morning, its index still on disk. To build that workspace faster the first time, start from a custom image that already has your JDK, Python or Node toolchain; sudo apt-get install -y openjdk-21-jdk works too, once per sandbox.

Mistakes and how Runtime handles them

• Choosing a password or a key file in Gateway. The installed entry's ProxyCommand is what reaches the sandbox, so choose the option that parses ~/.ssh/config.
• The IDE drops after a long break. An open connection does not keep a sandbox running; when its lease pauses or stops it, the session ends. Extend the lease, or make a paid sandbox --persistent for all-day work.
• Sessions past a day. Each SSH connection lasts at most 24 hours; open it again to go on.
• Several IDEs and forwards at once. An organization has at most 16 SSH logins and port forwards open together.
• Running out of disk. The default 4 GiB disk had about 2.5 GiB free on 24 September 2026, too little for an IDE backend and a large project; ask for more at create.

Gateway, VS Code or a shell?

Gateway suits Java, Kotlin and other projects where a JetBrains IDE's indexing and refactoring are the reason you use it. VS Code Remote-SSH keeps the editor on your machine and runs its server in the sandbox. For quick fixes, SSH in from a terminal. All of them use the same <id>.runtime host.

Related

• SSH and editors.
• Java in a sandbox and Kotlin.
• Per-user dev environments.

Sources

• JetBrains, "Connect and work with JetBrains Gateway", https://www.jetbrains.com/help/idea/remote-development-a.html, read 25 September 2026.
• JetBrains, "Prerequisites", https://www.jetbrains.com/help/idea/prerequisites.html, read 25 September 2026.

Facts on this page were checked on 25 September 2026.