find-skills

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow can invoke the Skills CLI to search/install skills from the open “skills ecosystem” (e.g., npx skills find / npx skills add), which may fetch and ingest outsider-authored skill package content (public web/GitHub) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill instructs using runtime package-install commands like "npx skills add owner/repo@skill" (e.g., "npx skills add vercel-labs/agent-skills@react-best-practices") which will fetch and execute remote code from GitHub/npm at runtime, so external content can directly execute code and control agent behavior.