fumadocs-article-importer

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md workflow explicitly fetches and ingests arbitrary public web pages via Jina MCP's read_url (Step 2) and then has the agent read and act on that untrusted content for concept extraction, cross-reference insertion, image/download decisions, classification and MDX generation—behavior that could allow indirect prompt injection from third‑party pages.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill calls external Jina endpoints at runtime (e.g., https://mcp.jina.ai/sse and the Jina API curl endpoint https://r.jina.ai/{article_url}) to fetch arbitrary article content that is then injected into AI prompts and relied on as a required dependency, so these runtime fetches can directly control the agent's input context.