SmolVM is an open-source local sandbox for AI agents on macOS and Linux.
I started building it because agent workflows need more than isolated code execution. They need a reusable environment: write files in one step, come back later, snapshot state, pause/resume, and increasingly interact with browsers or full desktop environments.
Right now SmolVM is a Python SDK and CLI focused on local developer experience.
Current features include: - local sandbox environments - macOS and Linux support - snapshotting - pause/resume - persistent environments across turns
Install: ``` curl -sSL https://celesto.ai/install.sh | bash smolvm ```
I’d love feedback from people building coding agents or computer-use agents. Interested in what feels missing, what feels clunky, and what you’d expect from a sandbox like this.
I'd like to use it for agent like OpenClaw, but i wanna know if agnet spawn sub agent in VM, and it's resource requirements is bigger than VM's resource, those processes killed by OOM? or SmolVM gracefully handle it? Or can i adjust resource limit for my purpose?
[dead]