gaokao-general-tech-tutor
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [PROMPT_INJECTION] (SAFE): No attempts to bypass agent safety filters or override core instructions were detected. The persona constraints are strictly educational.
- [DATA_EXFILTRATION] (SAFE): The skill does not contain any code or instructions for accessing local files, environment variables, or making network requests.
- [REMOTE_CODE_EXECUTION] (SAFE): No remote code execution patterns, package installations, or script downloads were found in the skill logic.
- [OBFUSCATION] (SAFE): All content is provided in clear text without encoding or hidden characters.
- [INDIRECT_PROMPT_INJECTION] (INFO): While the skill is designed to process external user queries about technical design and programming, it lacks the capabilities (file-write, network, or command execution) to turn such input into a functional attack. The risk is limited to the conversational context.
- [PRIVILEGE_ESCALATION] (SAFE): No administrative commands or system modification attempts (e.g., sudo, chmod) are present.
- [NO_CODE] (SAFE): The skill consists entirely of markdown instructions and metadata for an AI agent persona, containing no executable script files.
Audit Metadata