Security audit
Security checks for vulnerabilities and agentic risk
This skill is a simple local medical-abstract summarizer with some rough documentation, but no evidence of hidden, privileged, persistent, networked, or destructive behavior.
Before installing, treat this as a draft-quality helper for plain-language summaries, not a validated medical or regulatory compliance tool. Review generated summaries for accuracy, especially if used for patients, public communication, or EU CTR-related work.
This markdown file describes when to use the skill, but the activation conditions are generic and partly malformed, such as 'Use this skill when you need a documented fallback path' and repeated incomplete text. The section does not clearly distinguish this skill from many other academic-writing tasks or provide exclusion conditions, increasing the chance of unintended invocation.
The feature list specifies 'Reading level optimization (Grade 6-8)', which imposes a locale- and audience-specific language standard as a default behavior. The file does not indicate that this readability target is optional, user-selectable, or justified as a region-specific requirement.
The skill file identifies scripts/main.py as the primary implementation surface and instructs users to execute it, yet the risk table states "Network Access | No external API calls" as a concrete behavioral claim. In this file there is no code evidence supporting that claim, so the documentation asserts a specific non-network behavior without substantiation and in tension with the executable-based design.
No suspicious patterns detected.
