sync-skills

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly clones arbitrary GitHub repositories (sync_github_repo -> git clone) and fetches skillsmp.com pages via curl (sync_skillsmp_page -> curl -> /tmp/page.html) and then installs those user-provided files into ~/.agents/skills and other runtime skill directories, meaning untrusted, user-generated third-party content is ingested and can materially alter agent/tool behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The script performs runtime fetches of external skill content—e.g., "git clone https://github.com/user/repo.git" and "curl -sL https://skillsmp.com/skills/skill-name"—and then copies repository/parsed page files (including SKILL.md) into agent skill directories (~/.agents/skills), which lets fetched remote content directly provide or modify agent prompts/skills; therefore these URLs are runtime dependencies that can control agent behavior.