ai-elements

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

• Third-party content exposure detected (high risk: 0.80). The skill's documentation (e.g., references/agent.md and references/inline-citation.md) explicitly defines and shows tools like readUrl and webSearch and message parts of type "source-url" that accept arbitrary URLs and are intended to be fetched and parsed by the agent, meaning untrusted third‑party web content can be read and used to influence tool execution and decisions.