clean-web-markdown

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

• Third-party content exposure detected (high risk: 0.95). The required workflow fetches arbitrary web page content at runtime via the outsider-controlled URL (Cloudflare negotiation http_get(url, Accept: text/markdown, ...) and/or Jina Reader https://r.jina.ai/<url>), and that fetched free-form text is then inserted into the agent context as the returned markdown.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill performs runtime fetches of remote markdown that are injected into agent context — notably calls to the Jina reader service (https://r.jina.ai/) and the Firecrawl scrape API (https://api.firecrawl.dev/v1/scrape) — so those external endpoints can directly control the content given to the agent.