scan
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of local command-line tools including git and gh to manage repositories, diffs, and pull requests. It also supports execution of specific, whitelisted peer review CLI tools for cross-model analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as code diffs, pull request comments, and ticket descriptions. It incorporates strong defensive instructions, specifically directing the agent and subagents to treat this content strictly as data and to ignore any instructions embedded within them.
- [DYNAMIC_EXECUTION]: The skill features a proving mechanism that executes agent-generated test code to verify findings. It also supports running project-specific verification tools. These features are protected by security controls, including execution in isolated temporary directories and a requirement for explicit user trust flags before running project-specific code.
- [DATA_EXFILTRATION]: The skill can send code diffs to external models via peer review CLIs. This behavior is transparent, requiring an explicit user-provided argument and providing a mandatory disclosure statement before any data is transmitted.
Audit Metadata