crawl

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

• Third-party content exposure detected (high risk: 0.90). The skill calls the Tavily crawl API (POST https://api.tavily.com/crawl) to fetch arbitrary public URLs and returns raw_content or instruction-guided chunks from those sites (full pages, blogs, forums, etc.) which the agent is intended to read and use as context, exposing it to untrusted third-party content that could carry indirect prompt injections.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill makes runtime calls to https://api.tavily.com (e.g., POST https://api.tavily.com/crawl and /map) to fetch arbitrary page content which the skill then injects into LLM context/prompts, so the remote content can directly control agent prompts.