tavily

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

• Third-party content exposure detected (high risk: 1.00). The skill uses the Tavily web-search API and an extract script (node .../scripts/extract.mjs "https://example.com/article") to fetch and return content from arbitrary public URLs via Tavily (tavily.com), exposing the agent to untrusted third-party web content that it will read and interpret.