rubber-duck-review
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to use CLI tools like
codex,copilot, andclaudeto perform code reviews. It mandates the use of security-restricting flags such as--sandbox read-only,--deny-tool='write', and--permission-mode planto ensure secondary agents cannot alter the filesystem or execute arbitrary shell commands.\n- [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it processes codebase content through secondary agents. The skill implements the following security measures for this surface:\n - Ingestion points: The codebase is inspected directly by the secondary reviewer agents as described in
SKILL.md.\n - Boundary markers: The instructions require adding a clear disclaimer to the reviewer: 'This is a review-only task. Do not edit, write, or modify any files.'\n
- Capability inventory: The reviewer tools have access to project files but are restricted from writing or executing shell commands via CLI arguments.\n
- Sanitization: The skill relies on environment-level sandboxing and explicit instruction-based constraints to mitigate risk rather than content-level sanitization.
Audit Metadata