third-party-skills

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly lists and enables installation of skills like /reddit, /twitter, /producthunt, and /requesthunt that "search and retrieve" posts, tweets, comments and other public user-generated content from Reddit, X, Product Hunt and GitHub, which are untrusted third-party sources that the agent would read and could materially influence actions.