Vet AI agent skills for security vulnerabilities and permission scope.
npx skills add https://github.com/rongxinzy/RongxinAI --skill skill-vetter-rongxinzy
Please help me install this Agent Skill. Skill: skill-vetter Source: https://github.com/rongxinzy/RongxinAI/tree/main/SKILLs/skill-vetter Command: npx skills add https://github.com/rongxinzy/RongxinAI --skill skill-vetter-rongxinzy
This Skill helps identify potential security risks in AI agent skills before installation, ensuring safe and reliable skill deployment.
Run the following command to vet a skill from GitHub: curl -s "https://raw.githubusercontent.com/OWNER/REPO/main/skills/SKILL_NAME/SKILL.md" | skill-vetter
High-intent search queries and answers about installing and using this skill.
Vetting AI agent skills for security vulnerabilities involves scanning for malicious code, prompt injection, and data exfiltration. This assessment identifies red flags and harmful content to ensure safe skill deployment from untrusted sources.
Skill vetting is the security assessment of AI agent skills to identify malicious code and evaluate permission scope. It checks if the required file, network, and command access is appropriate for the skill's purpose.
To vet a skill from GitHub, run a shell command like `curl -s "[raw GitHub SKILL.md URL]" | skill-vetter`. This quickly gathers necessary repository information and assesses the skill's security and risk level.
You can classify security risks in AI agent skills into Low, Medium, High, and Extreme levels. This risk classification provides corresponding actions for each severity to guide safe deployment decisions.
The skill-vetter requires no external dependencies to scan for malicious code. It operates entirely within the scope of the SKILL.md file to identify suspicious patterns and harmful content.
kali20gakkiConvert GitHub blob links to raw URLs and fetch content via agent_router.
TencentClassify AI agent security findings using the OWASP ASI 2026 framework.
aniruddh2003Run static analysis tools to detect security vulnerabilities, type errors, and linting issues.
adaptive-enforcement-labScan code, dependencies, and containers for vulnerabilities and generate SBOMs.
