Run your coding agent

Prepare the workspace

Create a Machine, connect using the SSH column from ls, and clone your project inside it. Install dependencies as you would on a Linux computer. Use a separate Machine for unrelated users or untrusted projects. Agent credentials and project secrets saved in the Machine are part of its disk and Snapshots.

The agent runs in your Machine; requests to its model provider still leave Runin. Hosting the workspace in Germany does not move that provider's inference or account processing to Germany.

Claude Code

Follow the official Linux installation instructions inside the Machine. Start claude from the project directory and complete its sign-in flow using your browser. Keep the default permission checks while learning a new project.

Use tmux if the session should survive an SSH disconnect. Pause stops the Machine's processes; resume it before continuing work.

Codex

Install the Linux CLI using the official Codex guide. From the Machine, device-code login lets you approve sign-in in a browser on your own computer:

Your ChatGPT account or workspace must allow device-code login. See Codex authentication for supported alternatives. Then run codex from the project directory. Keep its approval and sandbox settings appropriate to the code you are running.

OpenCode

Install OpenCode inside the Machine using its official instructions, then select and authenticate your model provider:

Run opencode in your project. Provider billing and data processing follow the account you selected; Runin does not supply a model subscription.

Claude Managed Agents worker

Create a self-hosted environment in the Claude Console, then install Anthropic's ant CLI in a dedicated Machine using the self-hosted sandbox guide. Clone the required project and prepare its dependencies yourself.

Provide ANTHROPIC_ENVIRONMENT_ID and ANTHROPIC_ENVIRONMENT_KEY through a protected environment file. Do not put the key in a command argument, repository or shared Template. Start the worker in your project directory:

This simple worker executes work in the same Machine. Use it for one trusted project environment. For fresh isolation per session or attached memory stores, follow Anthropic's SDK worker instructions. Tool inputs and outputs still go to Anthropic. This is a setup guide; Runin does not yet provision or operate these workers for you.