interview
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including resumes, job descriptions (JD), project documents, and public links. These inputs are interpolated into the agent's logic to extract claims and generate interview questions, creating a potential attack surface where malicious instructions embedded in a resume or JD could influence the agent's behavior.
- Ingestion points: User-provided resume text, JD, project documents, papers, and public links (referenced in
SKILL.md). - Boundary markers: The skill does not explicitly instruct the agent to use specific delimiters or tags to isolate external content from its own instructions.
- Capability inventory: The skill reads local reference files (
references/*.md) and has the capability to write interview reviews to local files upon user request. - Sanitization: No explicit sanitization or filtering of external content is described in the provided instructions.
Audit Metadata