Fast hardware-isolated execution sandboxes for AI agents and untrusted code.
Features
- Hardware isolation
- About 30 ms P90 startup
- Disposable environments
- CLI and SDKs
- Claude plugins
- Example templates
Use Cases
- Agent code execution
- Untrusted code sandboxing
- Model evaluations
- Ephemeral development environments
FAQ
Fast hardware-isolated execution sandboxes for AI agents and untrusted code. Core capabilities include: Hardware isolation, About 30 ms P90 startup, Disposable environments.
Common scenarios include: Agent code execution, Untrusted code sandboxing, Model evaluations.
