Vetty is a security sandbox tool that runs untrusted code inside Firecracker micro-VMs. Security is a core concern of this project.

The following are considered in scope for security reports:

• VM escape — Guest code breaking out of the Firecracker VM boundary
• Host compromise — Guest code gaining unauthorized access to the host system
• Daemon vulnerabilities — Issues in the REST API, WebSocket, or vsock listener that could be exploited
• Network isolation bypass — Guest traffic bypassing NAT rules or accessing unintended host services
• Privilege escalation — Any path from unprivileged guest code to elevated host privileges

Please do NOT report security vulnerabilities through public GitHub issues.

Instead, please email: security@vetty.dev

Include:

• Description of the vulnerability
• Steps to reproduce
• Potential impact assessment
• Any suggested fixes (optional)

• Acknowledgment: Within 48 hours
• Initial assessment: Within 1 week
• Fix or mitigation: Depends on severity, typically within 30 days

Vetty is under active development and has not been formally audited. Known limitations:

1. The host network setup uses iptables rules managed via sudo — misconfiguration could weaken isolation
2. The mitmproxy HTTPS interception requires a CA certificate trusted by the guest, which is baked into the rootfs
3. Serial console access gives interactive shell access to the guest VM

Do not rely on Vetty as a sole security boundary for adversarial workloads in production environments.

• Run Firecracker with the recommended production host setup
• Keep Firecracker updated to the latest stable release
• Use a dedicated user for running Vetty with minimal privileges
• Review iptables rules after Vetty exits to ensure cleanup