jina

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly accepts arbitrary URLs and performs site-specific searches and web scraping (see SKILL.md and the JSON "url" input: "read and extract content from any URL"), so the agent will ingest untrusted public web content that could contain indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill accepts and fetches a user-supplied URL at runtime (e.g., "url": "https://example.com/article") and uses the fetched page content to drive extraction/search, so remote content directly controls the agent's prompts/context.