linkedin-interviewer
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is text-based interviewing and material organization. It does not utilize any tools for network communication, external package installation, or shell command execution.
- [INDIRECT_PROMPT_INJECTION]: The skill incorporates a specific section for 'Untrusted content' that instructs the agent to treat all user-provided or scraped data (such as LinkedIn bios or pasted text) strictly as data rather than instructions. This proactive design significantly mitigates the risk of indirect prompt injection.
- [DATA_EXPOSURE]: The skill manages user data by writing to local markdown files (
story-bank.md,voice-profile.md). The instructions include a specific rule to inform the user that these files should be excluded from version control (gitignored), ensuring that the raw interview material remains private to the user's environment.
Audit Metadata