tavily-best-practices

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

• Third-party content exposure detected (high risk: 0.90). The skill's documentation (SKILL.md and detailed references like references/search.md, references/extract.md, and references/crawl.md) explicitly instructs use of client.search(), client.extract(), client.crawl(), and client.research() to fetch and return raw content from arbitrary public URLs and web searches, and the integrations (LangChain/Anthropic/OpenAI examples) show agent workflows that consume those third‑party results to drive subsequent tool calls and synthesis—exposing the agent to untrusted third‑party content that can influence actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The Google ADK integration explicitly configures an MCPToolset to connect at runtime to https://mcp.tavily.com/mcp/, a required remote tool endpoint that supplies tool responses which are fed back into the agent (thereby directly influencing prompts/outputs), so this is a runtime external dependency that can control agent behavior.