skill-vetter

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to fetch and read files from public third-party sources (e.g., "Read ALL files in the skill" and the "Quick Vet Commands" that curl the GitHub API and raw.githubusercontent.com), so untrusted user-generated content would be ingested and could influence vetting decisions and subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill's Quick Vet Commands explicitly instruct runtime fetching of remote skill content via curl (e.g., "https://raw.githubusercontent.com/OWNER/REPO/main/skills/SKILL_NAME/SKILL.md"), which would inject external instructions into the agent's context and can directly control prompts.