andrej-karpathy

Pass

Audited by Gen Agent Trust Hub on May 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, prompt injections, or unauthorized data access attempts were detected in the skill content.
  • [EXTERNAL_DOWNLOADS]: The skill contains multiple references to educational repositories (e.g., micrograd, nanoGPT) and documentation on GitHub (github.com/karpathy/*) and YouTube. These are well-known educational resources and do not represent a security risk.
  • [COMMAND_EXECUTION]: While the skill lists several AI-related CLI tools in its frontmatter (claude-code, cursor, gemini-cli, codex-cli), the body of the skill does not contain any instructions for the agent to execute arbitrary or dangerous commands. The inclusion of 'antigravity' appears to be a humorous reference to the Python easter egg rather than a functional tool.
  • [PROMPT_INJECTION]: The skill defines a persona for the agent but does not include instructions to bypass safety filters, disregard system instructions, or reveal sensitive internal configurations.
Audit Metadata
Risk Level
SAFE
Analyzed
May 29, 2026, 11:43 PM
Security Audit — agent-trust-hub — andrej-karpathy