system-architect

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is purely instructional, providing a framework for identifying system boundaries and generating architectural documentation. It does not perform any file system modifications (other than suggesting documentation paths), network requests, or remote code execution.
  • [PROMPT_INJECTION]: The instructions follow standard persona-based prompting and do not contain any attempts to bypass safety filters, override system prompts, or extract internal instructions.
  • [DATA_EXPOSURE]: No hardcoded credentials, API keys, or sensitive file paths (such as .ssh or .aws) were detected. The file paths mentioned (e.g., genesis/v{N}/...) are project-specific documentation paths.
  • [REMOTE_CODE_EXECUTION]: There are no patterns involving remote script downloads (curl | bash) or package installations (npm/pip). The bash blocks in the documentation are illustrative examples of agent-internal slash commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 06:59 AM
Security Audit — agent-trust-hub — system-architect