react-quiz
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION] (CRITICAL): The skill directs the agent to execute untrusted code provided by the user using Node.js. Ingestion point: User input for 'Type D: Write Code' questions in SKILL.md. Capability inventory: Execution of Node.js and tsc (TypeScript compiler) subprocesses. Sanitization: None provided. Boundary markers: Absent. This configuration allows an attacker to gain full control over the environment by submitting malicious payloads as 'answers'.
- [COMMAND_EXECUTION] (CRITICAL): The requirement to use system-level interpreters (Node.js) on user-controlled input without sandboxing is a severe security flaw.
- [PROMPT_INJECTION] (HIGH): The skill is susceptible to Indirect Prompt Injection (Category 8) because it ingests untrusted external data and processes it using high-privilege tools (the Node.js runtime).
Recommendations
- AI detected serious security threats
Audit Metadata