jina-reader

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

• Third-party content exposure detected (high risk: 0.95). The required workflow fetches public web content at runtime via the helper calling https://r.jina.ai/ with a user-supplied URL, and the returned Markdown is then printed/ingested into the agent’s LLM context—this is outsider-authored free text from arbitrary third-party pages (including X/Twitter pages).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime helper (scripts/read_url.py) performs GET/POST requests to https://r.jina.ai/ (e.g., "https://r.jina.ai/" or POST to "https://r.jina.ai/") and returns remote Markdown that is injected into the agent's context, meaning external content from that URL can directly influence prompts/behavior.