Skip to content
CommandChain CommandChain CommandChain Docs

Connect an AI account for managed runs

You only need this page if your workspace runs work on CommandChain’s managed servers. A runner on your own machine signs in for itself and uploads that sign-in only if you choose to push it for managed runs. See install a runner.

When work runs on CommandChain servers

Section titled “When work runs on CommandChain servers”

Managed runs happen on machines we operate, so they need a coding-agent login uploaded to your workspace. You would use them when:

  • You are on Windows. Self-hosted runners are Linux and macOS only.
  • You would rather not install anything, or want work to keep moving while your laptop is closed.

In the setup wizard’s last phase, an admin selects Run on CommandChain servers instead. That turns on the workspace setting Allow work to run on CommandChain servers (Settings, then Budgets), which you can change later. See set budgets.

Connect your AI account for managed runs

Section titled “Connect your AI account for managed runs”

With managed runs on and no runner of your own online, the wizard shows Connect your AI account for managed runs, and waits for a login to arrive.

On a Mac or Linux machine where the runner is installed and paired and you are signed in to that agent, open the runner with commandchain-runner and press c, or run commandchain-runner creds push --agent claude-code.

Terminal window
commandchain-runner creds push --agent claude-code
commandchain-runner creds push --agent codex
commandchain-runner creds push --agent gemini

This uploads your personal sign-in for that coding agent, for runs on CommandChain’s managed servers. It is a different login from the runner’s own sign-in, which the runner uploads only if you choose to push that one too, by adding --from-runner-signin. Uploaded logins are listed as metadata only, never as token material.

Back in the wizard, the last check turns green on its own. Check again forces a refresh if you are impatient, and Use my own machine instead goes back to installing a runner.

See managed runner credentials for what is uploaded, how to list it, and how to remove it.

The other two ways an agent can think

Section titled “The other two ways an agent can think”

A runner, on your machine or ours, is the default. Two others exist and you can mix them:

  • Trial credits. In the cloud beta, the platform funds your team’s thinking up to a granted balance. Nothing to enter and nothing to install: it is already there on day one, which is what lets the kickoff huddle happen before you have a runner. See trial credits.
  • Your own provider key (Direct-LLM). Open Settings, then AI Backend, choose Direct LLM, and enter your Anthropic, OpenAI, or Gemini key. Your agents then call that provider directly and usage bills to your provider account. The backend is a per-workspace setting with an optional per-agent override, so you can put most of your team on a runner and pin one agent to Direct-LLM, or the reverse.

Next steps

Section titled “Next steps”