oma-qa
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to ingest and analyze untrusted codebase content such as diffs, PRs, and build outputs.
- Ingestion points: Codebase files, pull request diffs, and build outputs as defined in SKILL.md.
- Boundary markers: No specific delimiters or safety instructions for delimiting processed data are defined in the instructions.
- Capability inventory: The skill has significant capabilities including executing shell commands (npm audit, bandit, lighthouse) and comprehensive browser automation (click, fill, evaluate_script) for functional verification.
- Sanitization: The instructions do not specify sanitization or escaping of processed code before analysis.
- [COMMAND_EXECUTION]: The skill performs legitimate command execution for diagnostic and testing purposes.
- Evidence: The skill is instructed to run npm audit, bandit, and lighthouse, and to execute project-specific start and test commands during the runtime verification phase.
- [DYNAMIC_EXECUTION]: The skill utilizes dynamic execution in a browser context to verify application state during quality reviews.
- Evidence: The execution protocol includes the use of evaluate_script within a browser environment to inspect DOM state and global variables.
Audit Metadata