wecomcli-pptx
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external documents (PPTX, DOCX, and PDF). To defend against indirect prompt injection, the instructions mandate the neutralization of executable tags like and , as well as all HTML event handlers, before data is written to any document.
- [COMMAND_EXECUTION]: Local file processing and interaction with the wecom-cli tool are managed through specific Python scripts and command-line interfaces. Security is maintained through a strict 'Command Splicing Security Specification' that enforces character escaping and ensures that all external inputs are handled as pure literals to prevent shell injection.
- [DYNAMIC_EXECUTION]: The doc_image_extractor.py script employs import for checking optional dependencies, and pptx2image.py uses subprocess.run (without shell=True) to invoke LibreOffice for PDF conversion. These dynamic operations are functionally integral to the skill and are constrained by safety guidelines.
- [EXTERNAL_DOWNLOADS]: The skill interacts with the well-known Kroki service for diagram rendering and utilizes official WeCom infrastructure for image management and document synchronization. These external interactions are disclosed and align with the skill's intended use case.
Audit Metadata