tavily-mcp

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to call Tavily MCP tools that fetch and ingest public web content (e.g., Workflow 2's tavily_extract for user-provided URLs, Workflow 3's tavily_crawl/tavily_map for crawling sites, and Workflow 4's tavily_research for multi-source synthesis), so the agent will read untrusted third-party/user-generated pages whose content can influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly calls runtime fetch tools like tavily_extract/tavily_crawl that retrieve and inject user-supplied web pages (e.g., https://example.com/article) into the agent's context, meaning remote page content can directly control prompts and responses.