check-studio-skills

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs the agent to use query_documents/get_document to fetch _type == "skill" documents from a target Sanity dataset (runtime external Studio content), and then to read the content markdown and follow it—this is outsider-authored free text unless the dataset content is guaranteed to be authored by the operating user/organization.