interview-system-designer

Warn

Audited by Gen Agent Trust Hub on Apr 5, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: There is a fundamental contradiction between README.md and SKILL.md. The README focuses on technical system design interviews, while the SKILL instructions focus on HR hiring processes and interview loop planning. This discrepancy functions as metadata poisoning and can lead to model confusion or unexpected behavior.
  • [COMMAND_EXECUTION]: The skill instructions in SKILL.md direct the agent to execute a local script (scripts/interview_planner.py) via the command line. It explicitly passes user-provided input (--role "<role>") as an argument. This pattern is vulnerable to command injection if the underlying script or the agent's shell execution environment does not properly sanitize the input.
  • [NO_CODE]: The skill relies on several referenced files, including scripts/interview_planner.py and multiple markdown files in a references/ directory. None of these files are included in the skill package, making the script's actual logic and safety unverifiable.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 5, 2026, 08:05 PM
Security Audit — agent-trust-hub — interview-system-designer