npx-skills

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow reads the project’s checked-in skills-lock.json and then installs skills from the specified GitHub source values; those installed skill files are outsider-authored repository content that becomes LLM-readable text in .claude/skills/ (indirect prompt injection risk via third-party skill code/content).

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

• Potentially malicious external URL detected (high risk: 0.90). The CLI explicitly installs agent skills at runtime from GitHub sources (e.g., https://github.com/vercel-labs/skills and arbitrary <owner/repo> GitHub URLs used with npx skills add), which fetches remote skill files that directly control agent prompts and are required for operation.