meta-security-review-bundle
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a defensive tool designed to improve security posture. It coordinates multiple independent security sub-agents to evaluate code changes or system operations.
- [PROMPT_INJECTION]: The skill manages a surface for indirect prompt injection when evaluating user-provided operations. * Ingestion points: User input enters the agent context via the inputs.user_message variable in steps policy_review, secret_scan, and audit_emit of SKILL.md. * Boundary markers: The task prompts utilize triple-dash separators to isolate user-controlled data from instructions. * Capability inventory: The skill metadata declares access to filesystem-read, network-read, and subprocess tools. * Sanitization: The skill applies xml_escape and truncation to user input to prevent it from escaping the designated data block or overriding the sub-agent's instructions.
Audit Metadata