mirrord-operator
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements excellent security practices for infrastructure management.
- Command Execution: All cluster-modifying commands (e.g., helm, kubectl) require explicit user review and approval before execution, preventing unauthorized or automated changes.
- Input Sanitization: The skill defines clear rules for validating user-supplied inputs like namespaces and pod names using regex, and it explicitly rejects shell metacharacters to prevent command injection.
- Credential Protection: It strictly forbids the transmission of license keys or secrets via command-line arguments, recommending the use of Kubernetes Secrets and keyRef references in values files instead.
- Data Ingestion: While the skill processes untrusted user configuration, it mitigates potential indirect prompt injection by using boundary markers (<USER_INPUT>) and instructing the agent to treat the content as opaque data.
- Trusted Sources: All external references and documentation links point to the official domains of the vendor (metalbear.com, mirrord.dev) or their official GitHub repositories.
Audit Metadata