glmv-doc-based-writing
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill transmits document content and the user's API key to the official Zhipu AI API (open.bigmodel.cn). This is the primary function of the skill and uses the vendor's official infrastructure.\n- [EXTERNAL_DOWNLOADS]: The script allows the agent to fetch and process documents from remote URLs provided by the user during execution.\n- [COMMAND_EXECUTION]: The core logic is implemented in a Python script (scripts/doc_based_writing.py) that is executed to handle document parsing, image conversion, and API communication.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from external documents, which is a surface for indirect prompt injection. 1. Ingestion points: File URLs or local paths provided to scripts/doc_based_writing.py. 2. Boundary markers: None. 3. Capability inventory: Network requests via the requests library and file system read access via PyMuPDF. 4. Sanitization: None. This is documented as a standard functional risk for document-processing skills.
Audit Metadata