morning-briefing

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

• Third-party content exposure detected (high risk: 0.80). The skill's Step 4 explicitly runs web_search queries to pull news from the open web (and Step 3 uses Gmail to fetch user emails), and those untrusted third-party items are read and summarized as part of the mandatory briefing composition/workflow, so external content could indirectly inject instructions or influence the agent's outputs and decisions.