skills/lijigang/ljg-skills/ljg-teach/Gen Agent Trust Hub

ljg-teach

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The instructions focus entirely on a specific teaching methodology and do not contain patterns designed to bypass AI safety filters or override system instructions.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or network exfiltration patterns were detected. The skill does not request or use tools that interact with the local filesystem or external network.
  • [OBFUSCATION]: The content is provided in clear text with no evidence of Base64 encoding, zero-width characters, or homoglyph-based evasion techniques.
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill does not include any package installation commands, remote script downloads, or dynamic code execution patterns.
  • [PRIVILEGE_ESCALATION]: No commands related to privilege escalation (e.g., sudo, chmod) or persistent system modifications were found.
  • [INDIRECT_PROMPT_INJECTION]: While the skill ingests user-provided concepts for teaching, it possesses no exploitable capabilities (such as shell access or web browsing) that could be leveraged if malicious instructions were embedded in the user input.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 01:14 AM
Security Audit — agent-trust-hub — ljg-teach