lyxy-document-reader

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly documents support for "HTML / URL (网页内容)" and shows the runtime command python scripts/lyxy_document_reader.py <文件路径或URL>, indicating the skill will fetch and parse arbitrary web pages (untrusted third‑party content) which the agent ingests and could materially influence subsequent parsing or actions, enabling indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.80). The SKILL.md explicitly accepts and fetches user-provided HTTP(S) URLs at runtime (user-supplied "http://"/"https://..." URLs), and those fetched page contents are parsed and injected into the agent's output context — effectively allowing arbitrary external content to influence or control prompts/responses.