Skip to main content
Early access. Cloud sessions and managed agent runs are not open to every account yet: we are letting accounts in gradually while we test. They run in a cloud sandbox operated by Mutagent, so there is nothing to host. A sandbox with nothing to do for 15 minutes stops; send an interactive session a message and it wakes up, delivers your message and carries on in the same conversation. A -p task run cannot be woken; run it again.
The platform runs agents in cloud sandboxes, with the LLM providers and Environments of your workspace. There are two ways to use it: Both are driven from the mutagent CLI and the SDKs.

A managed agent in three commands

These assume the invoice/ folder from the quickstart, the CLI, Bun 1.3.14, and a sign-in with a workspace selected. The quickstart’s checklist has the details.
check validates the folder on your machine. deploy uploads it as revision v1 and makes it active in the default slot (no --env). mutagent helix agent @<slug> runs the active revision in a cloud sandbox and prints the answer. Each command exits 0 on success; add --json to check and deploy for a result a script can read. The quickstart builds this agent step by step.

Managed agents

What a managed agent is: revisions, slots and runs.

Quickstart

Write, check, deploy and run your first managed agent.

Helix Cloud

Run Helix on a task or as a session in a cloud sandbox.

Cloud sandboxes

What the platform manages for each run.