Do NOT open a public issue for security vulnerabilities.

Instead, please report security issues via:

1. Email: security@pentest-labs.io
2. GitHub Security Advisories: Create a private advisory

• Description of the vulnerability
• Steps to reproduce
• Potential impact
• Suggested fix (if any)

• Acknowledgment: Within 48 hours
• Initial Assessment: Within 7 days
• Fix Release: Within 30 days for critical issues

This platform runs untrusted code in containers. Security is multi-layered:

• No privileged containers
• All capabilities dropped (cap_drop: ALL)
• no-new-privileges enforced
• Seccomp profiles applied
• No host mounts from user templates
• No Docker socket access for zone containers

• Each zone gets an isolated Docker bridge network
• No inter-zone communication possible
• Internet egress denied by default
• Only declared ingress ports exposed via Traefik

• CPU, memory, PID limits per container
• Disk quotas per zone
• Zone TTL enforcement (auto-destroy)
• Maximum concurrent zones per user

• JWT authentication with short-lived tokens
• Rate limiting at Traefik level (auth: 10/min, API: 100/min, AI: 40/hr)
• Input validation on all endpoints
• CORS restricted to configured origins

• Flag values never sent to AI providers
• Output filtering for flag patterns
• Rate limiting per user per zone
• All AI conversations logged for audit

• Schema validation on import
• Image allowlist enforcement
• No privileged container declarations allowed
• No host mount declarations allowed
• Admin approval required for new templates

• Zone containers share the host kernel (container escape is a theoretical risk)
• Terminal sessions are not encrypted end-to-end (rely on HTTPS transport)
• AI flag-leak prevention is heuristic-based (not cryptographically guaranteed)

We follow responsible disclosure practices. Reporters will be credited in the changelog unless they prefer anonymity.