om-qa-buddy
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from pull requests, issue descriptions, and comments. It includes explicit defensive instructions for the agent to treat this content as data rather than instructions, specifically directing the agent to disregard and report any embedded commands or behavior overrides.\n- [COMMAND_EXECUTION]: The skill uses shell commands for git operations and environment management. It mitigates injection risks by validating externally-sourced identifiers against strict regular expressions before shell interpolation.\n- [DYNAMIC_EXECUTION]: The skill generates a local interactive HTML runbook containing JavaScript for tracking progress. Untrusted strings from the QA process are interpolated into this template. This creates a potential surface for local cross-site scripting (XSS) if data is not correctly escaped during the file creation process.\n- [CREDENTIALS_UNSAFE]: The skill handles authentication credentials by referencing them through environment variables loaded from configuration files, ensuring that raw secrets are not included in the model's output, logs, or screenshots.
Audit Metadata