competitor-teardown

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

• Third-party content exposure detected (high risk: 0.86). The required workflow uses runtime web research and scraping via commands like belt app run tavily/search-assistant and belt app run exa/search/tavily/extract (and browser screenshots via infsh/agent-browser), which necessarily ingests outsider-authored free-form text from third-party websites into the agent’s LLM context.