creating-agent-skills

Pass

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The instructions include specific safety guardrails for handling untrusted content, such as treating outside skills as text and requiring a sandbox for code execution.
  • [SAFE]: The skill mandates explicit user approval for any destructive actions, external writes, or significant increases in scope.
  • [SAFE]: Analysis of the instructions and reference materials found no evidence of prompt injection, data exfiltration, or hidden malicious code.
  • [SAFE]: The scoring rubric included in the reference materials encourages the evaluation of safety boundaries and authorization controls in other skills, promoting secure development practices.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 8, 2026, 12:13 PM
Security Audit — agent-trust-hub — creating-agent-skills