rails-principles-and-boundaries

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No evidence of prompt injection or instructions to bypass safety guidelines was found. The skill uses natural instructional language to guide the agent on Rails development practices.
  • [COMMAND_EXECUTION]: The skill instructs the agent to detect and run project-specific linting commands (e.g., bundle exec rubocop, npm run lint). These are standard development tools and the skill specifically warns the agent not to assume a specific linter but to check the project's own configuration first.
  • [DATA_EXFILTRATION]: No network operations or sensitive file exfiltration patterns were detected. The skill mentions sensitive data contexts like PII and IDOR only to provide guidance on security reviews.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns, external downloads, or piped shell commands were found.
  • [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private tokens are present in the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 12:30 AM