FAQ
Answers to the most common questions about StateSet Sandbox.
What is StateSet Sandbox?
It is a production runtime for AI agents that provides isolated code execution, file I/O, and artifact storage on top of Kubernetes.
How is this different from running containers myself?
We handle isolation, lifecycle, timeouts, limits, and a stable API/SDK so agents can safely run code without you managing workers or custom orchestration.
How are files read and written?
File content is base64 encoded over the API and decoded inside the sandbox. This avoids binary corruption and keeps file I/O reliable for agent workflows.
What are the file size limits?
Default limits are 10MB per file and up to 100 files per request. Large artifacts should be uploaded via the artifact API.
How long do sandboxes run?
Each sandbox has a timeout (default and plan-based). You can set a custom timeout per sandbox within your plan limits.
Is the system isolated?
Each sandbox runs in its own Kubernetes pod with strict file path validation and non-root execution.
Does it work with Claude Code?
Yes. The sandbox image includes Claude Code CLI, and BYOK makes it easy to provide your Anthropic key.
Where can I deploy?
You can use the hosted API or run the controller in your own Kubernetes cluster.
Still have questions?
Sign up to explore the dashboard or review pricing details.