_project-context

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly supports adding knowledge sources from websites (see "Available Skills" -> /copilot-studio:add-knowledge with "website") and adding generative nodes that search and summarize content ("/copilot-studio:add-generative-answers" for SearchAndSummarizeContent or AnswerQuestionWithAI) which indicates the agent will ingest and act on untrusted public web content from the knowledge/ sources, enabling indirect prompt injection.