agent-workflow-patterns

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

• Third-party content exposure detected (high risk: 0.90). The researchAgent template in templates/react-agent.ts defines a "search" tool (noted as web scraping/API) and a "readArticle" tool that accepts arbitrary URLs and returns article content, so the agent will fetch and interpret untrusted public web content (arbitrary web pages) that could carry indirect prompt injection.