ocr-and-documents
Pass
Audited by Gen Agent Trust Hub on Jun 29, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads AI models and dependencies required for OCR functionality.
- The
marker-pdflibrary downloads approximately 2.5GB of pre-trained models from Hugging Face to the local user cache (~/.cache/huggingface/) upon first use. - The documentation and scripts (
scripts/extract_marker.py) proactively check for sufficient disk space (5GB) before proceeding with installation or execution, which is a defensive best practice. - [COMMAND_EXECUTION]: The skill provides various shell commands for document processing.
- Commands involve standard library installations (
pip install) and execution of local helper scripts for PDF and document parsing. - These operations are limited to the primary task of document extraction and do not involve unsanitized user input or privilege escalation.
- [REMOTE_CODE_EXECUTION]: While the skill suggests using
web_extractfor remote URLs, this is a standard tool invocation for remote data fetching rather than code execution. No patterns of piping remote content to a shell (e.g.,curl | bash) were detected.
Audit Metadata