oma-refactor
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
resources/hotspots.pyscript executes Git commands and thelizardtool usingsubprocess.run. Evidence: The script uses list-based arguments for subprocess calls (e.g.,subprocess.run(prefix + ["log", ...])), avoiding shell injection risks. It also implements path validation usingpathlib.Path.is_relative_toto ensure file operations remain within the repository boundaries. - [EXTERNAL_DOWNLOADS]: The documentation suggests using
uvxto execute tools likelizardfrom external registries. Evidence:resources/measurement.mdcontains the commanduvx lizard -l <lang> --CCN 10 <path>, which facilitates the dynamic download and execution of packages from the Python Package Index (PyPI). This is a standard developer workflow and is documented as a measurement option. - [INDIRECT_PROMPT_INJECTION]: The skill processes external codebase content and git logs which could theoretically contain malicious instructions. Ingestion points: Target source files, git commit history, and lint/metric tool outputs. Capability inventory: File-writing (code and tests), command execution (git, lizard, radon, test runners). Sanitization:
hotspots.pyuses NUL-separated git outputs and path validation to handle untrusted filenames safely.
Audit Metadata