evaluating-candidates
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE] (SAFE): The skill is comprised solely of Markdown documentation, templates, and reference guides. There are no Python scripts, JavaScript files, shell scripts, or binaries included.
- [DATA_EXFILTRATION] (SAFE): No network-related commands (curl, wget, etc.) or file-system-writing operations were found. The skill contains explicit instructions to avoid requesting secrets and to use redacted excerpts for sensitive candidate information.
- [REMOTE_CODE_EXECUTION] (SAFE): There are no mechanisms for downloading or executing remote code. All workflow steps are handled via natural language processing within the agent's context.
- [PROMPT_INJECTION] (SAFE): The instructions follow standard operational guidelines for hiring. No adversarial patterns like 'ignore previous instructions' or bypass attempts were detected.
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill is designed to process untrusted data (resumes, portfolios, and interview notes). While this represents an ingestion surface, the total absence of executable capabilities (no shell, no network, no file-write) ensures that any malicious instructions within candidate data cannot be exploited by the skill.
Audit Metadata