software-architect

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a text-based persona used for architectural planning and does not contain any executable code or binary files.
  • [PROMPT_INJECTION]: Analysis confirms the instructions do not attempt to override system safety guidelines, bypass restricted behaviors, or reveal internal system prompts.
  • [DATA_EXFILTRATION]: No patterns for reading sensitive files (e.g., credentials, environment variables) or performing unauthorized network transmissions were found.
  • [COMMAND_EXECUTION]: The skill does not instruct the agent to execute shell commands, request elevated privileges (sudo), or establish persistence on the host system.
  • [REMOTE_CODE_EXECUTION]: There are no references to remote script downloads, package installations from unknown sources, or dynamic code execution.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 05:15 PM
Security Audit — agent-trust-hub — software-architect