pdf-processing-pro
Pass
Audited by Gen Agent Trust Hub on Mar 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The analyzed script analyze_form.py demonstrates safe programming practices, including input validation and structured error handling.
- [EXTERNAL_DOWNLOADS]: The documentation references standard Python packages available on PyPI and provides links to well-known community resources for the Tesseract-OCR binary.
- [COMMAND_EXECUTION]: Workflow examples illustrate calling local Python scripts via the CLI and orchestrating them using subprocess.run for batch processing tasks.
- [PROMPT_INJECTION]: The skill acts as a data ingestion tool for PDF files, which introduces an inherent surface for indirect prompt injection where text inside a processed document could contain instructions for the agent.
- Ingestion points: PDF files processed by scripts such as analyze_form.py and extract_text.py.
- Boundary markers: Absent. Extracted text is handled as raw data without specific delimiters to prevent agent obedience to embedded instructions.
- Capability inventory: File system read/write operations for PDF, JSON, and CSV formats.
- Sanitization: Absent. The skill does not sanitize extracted content for potential prompt injection strings.
Audit Metadata