web-search-verifier

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to call the mcp_websearch_web_search_exa tool to fetch and extract "original_text" from public third-party websites (e.g., tradingeconomics.com, bloomberg.com, finance.yahoo.com) and requires using those fetched texts to determine verified numeric values, so untrusted web content can directly influence decisions and outputs.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). This skill requires the agent at runtime to fetch and verbatim-inject external webpage text as "original_text" (e.g., https://tradingeconomics.com/united-states/stock-market) into the model's output/verification context, meaning remote content can directly influence/alter the agent's prompts or decisions.