evaluating-trade-offs

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • NO_CODE (SAFE): The skill is composed strictly of Markdown documentation, templates, and instructions. It does not include any scripts (Python, JS, Shell), binaries, or executable configuration files.
  • EXTERNAL_DOWNLOADS (SAFE): The metadata in skillpack.json includes reference URLs to the original source at refoundai.com. These are informational links and are not used by the skill to download or execute remote code.
  • PROMPT_INJECTION (SAFE): The instructions follow a standard task-oriented structure without any attempts to bypass safety filters or override system prompts.
  • DATA_EXFILTRATION (SAFE): There are no network-capable tools or functions defined in this skill that could be used to exfiltrate data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 03:51 PM