discover-skills

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

• Third-party content exposure detected (high risk: 0.90). The skill's Web Researcher and Registry-Scout teammates (see references/team-templates.md and the SKILL.md "web-researcher" / "registry-scout" sections) explicitly fetch and extract content from public sources (brave-search, GitHub, Reddit, Hacker News, blogs, skills.sh) and those findings are used to rank candidates and drive install/create decisions, so untrusted third-party content is read and can materially influence agent actions.