dojo-jensen-huang
Pass
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of markdown documentation and instructional text. There are no executable scripts (.sh, .py, .js), configuration files for tool access, or external network requests.
- [NO_CODE]: No code or external packages were detected. The analysis confirms the absence of automated behaviors beyond standard text generation based on the provided persona context.
- [PROMPT_INJECTION]: Instructions found in
SKILL.mdandpersona.mdare designed to constrain the AI's tone and reasoning to match the Jensen Huang persona. No evidence of malicious bypasses, 'jailbreaks', or attempts to extract system prompts was found. - [DATA_EXFILTRATION]: No sensitive file paths, credential patterns, or network communication tools (like curl or wget) are present.
- [OBFUSCATION]: A manual and automated scan of all 21 files revealed no Base64 encoding, zero-width characters, homoglyphs, or other obfuscation techniques.
Audit Metadata