qiaomu-markdown-proxy

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests arbitrary public URLs (see SKILL.md "Triggers on any URL..." and the proxy priority entries for r.jina.ai / defuddle.md and the built-in scripts fetch_weixin.py and fetch_feishu.py), and then requires the agent to read/interpret and act on that content (display, summarize, save), so untrusted third‑party content can materially influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill runtime explicitly curls user-supplied URLs via the r.jina.ai proxy (curl -sL "https://r.jina.ai/{original_url}") and falls back to defuddle.md (curl -sL "https://defuddle.md/{original_url}"), injecting the fetched Markdown into the agent response/context — meaning remote content can directly control prompts/instructions at runtime.