Path Traversal Security Test
Simulates path traversal attempts to test sandbox isolation and sensitive file protections in agent runtimes.
Path Traversal Security Test is an automated test fixture designed to evaluate security boundaries within AI skill managers and agent execution environments. By attempting to read files outside its designated store—such as relative secrets files and sensitive system paths like /etc/passwd—it helps developers verify sandbox restrictions, test path-traversal mitigations, and ensure that AI agents cannot inadvertently expose restricted local files.
主要功能
01Simulates directory traversal attempts outside authorized skill directories
02Verifies package manager boundary enforcement and file lock integrity
03Tests AI agent runtime restrictions and file-access isolation policies
040 GitHub stars
05Provides reproducible attack vectors for unit and integration testing
使用场景
01Benchmarking vulnerability scanning tools for AI agent capabilities
02Testing sandbox isolation in AI skill runtimes and package managers
03Verifying automated defenses against malicious or insecure skill packages