glmv-doc-based-writing

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly accepts arbitrary document URLs ("Document (URL)" in SKILL.md and the --files CLI parameter) and the script (scripts/doc_based_writing.py) appends each remote URL as a "file_url" sent to the GLM-V model, meaning untrusted third‑party documents can be read/interpreted by the model and thus could inject instructions that materially influence outputs.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill accepts and sends user-supplied remote document URLs (e.g., https://example.com/doucment1.pdf) to the GLM-V API as file_url entries at runtime, so fetched remote content is injected into the model's input and can directly control prompts/instructions, which the skill requires to operate.