Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only prompt-engineering skill with one cautionary refusal-reframing example, but no hidden execution, persistence, credential handling, or data exfiltration behavior.
Reasonable to install if you want prompt-engineering templates and troubleshooting guidance. Review the refusal-related troubleshooting example critically, avoid using it to bypass safety rules, and adapt platform-specific examples such as file paths to your environment.
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
- Explain methodology before results
### Don'ts
- Don't make predictions beyond 12 months without caveats
- Don't ignore outliers without investigation
- Don't present correlation as causation
- Don't use jargon without explanation
The troubleshooting guidance recommends reframing a harmful request with fictional or novel-writing context to get past a model refusal. Even though presented as a prompt-engineering tip, this normalizes jailbreak-style refusal circumvention and can help users obtain assistance for unsafe topics under a benign pretext.
The file includes natural-language instructions that require analysis to always follow an English output structure such as 'Executive Summary', 'Key Findings', and 'Recommendations'. This imposes a language/locale expectation without any opt-in, alternative locale support, or justification that the skill is region-specific.
The guidance explicitly marks Windows-style paths as incorrect and recommends only Unix-style alternatives. This imposes a platform-specific convention in natural language without offering user opt-in or documenting a justified scope limitation, which can violate locale/environment flexibility expectations.
No suspicious patterns detected.