agentic-ai-research-agent

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

• Third-party content exposure detected (high risk: 0.85). The runtime workflow uses the tavily_search_tool (public web search results) and wikipedia_search_tool (Wikipedia page content/summary) whose returned text is appended into the LLM messages as {"role":"tool","content": str(result)}, creating an indirect prompt-injection channel from outsider-authored web content.