skills-sh-search

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

• Third-party content exposure detected (high risk: 0.75). El flujo “Explorar web” usa WebFetch para leer contenido legible de páginas públicas de skills.sh (p. ej., https://skills.sh///), que es texto de terceros y puede incluir prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs runtime fetching of external content — e.g., WebFetch to https://skills.sh/// (to load a SKILL.md that can inject instructions) and npx commands like npx skills add https://github.com/<owner>/<repo>/npx skills find which pull and execute remote code — so external content fetched at runtime can directly control prompts or execute code.