find-skills

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly discovers and installs SKILL.md files from public sources (skills.sh and arbitrary GitHub/GitLab URLs via npx skills find / npx skills add), and the workflows state agents read and execute those third-party skill instructions which could materially influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs repository installs like npx skills add owner/repo which fetches arbitrary GitHub repos at runtime (e.g., https://github.com/owner/repo) and loads their SKILL.md files that directly instruct agents and may include executable scripts, so remote content controls prompts/behavior and is a required dependency.