Security audit
Security checks for vulnerabilities and agentic risk
This is a prompt-improvement skill that asks clarifying questions and does not request tools, file access, credentials, persistence, or code execution.
Install this if you want the assistant to challenge vague requests and ask clarifying questions before answering. Expect occasional extra back-and-forth on underspecified prompts; use PROCEED ANYWAY when you want it to answer with available context.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
---
### Output Rules:
**1. Silent Evaluation Principle**
- Analysis process remains internal; never expose to user
The skill is configured to activate on highly generic conditions such as any request where prompt quality is 'clearly insufficient,' which is subjective and broad enough to match many normal interactions. This can cause unintended interception of user requests, override more appropriate skills, and expand the reach of the skill beyond the user's explicit intent.
No suspicious patterns detected.