zj-storage-analyzer
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local system utilities such as
du,sw_vers,uname, anddiskutil(on macOS) or standard filesystem APIs (on Windows) to gather disk usage metadata. It also usesosascriptandshell32.SHFileOperationWto manage file deletion and trashing. All commands are executed using safe list-based arguments without spawning a shell, minimizing injection risks. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of directory and file names encountered during disk scanning. While this could potentially be used to influence the agent's categorization logic, the skill provides strict guidelines for the agent to follow and requires manual user confirmation through a local web interface for any destructive actions.
- [DYNAMIC_EXECUTION]: The
server.pyscript launches a local web server (127.0.0.1) to serve the interactive report and provide a file management API. This server uses a randomized session token, restricts requests to the local host to prevent DNS rebinding, and validates all file paths against a strict allowlist and directory boundaries ($HOME) before performing any actions.
Audit Metadata