test-lab

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The repository exposes a blind-eval HTTP endpoint (service.evaluate) that accepts client-supplied "blind_tests" strings and executes them as Python code or shell commands inside the container (with mounted workspaces), enabling remote code execution and straightforward data exfiltration of mounted files or environment secrets.