extract-slide-text
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
extract_slide_text.pyscript executes thepdftotextbinary usingsubprocess.run. The command is constructed using a list of arguments withoutshell=True, which prevents shell command injection.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external PDF files, which presents a surface for indirect prompt injection attacks.\n - Ingestion points: The script reads PDF data through the
pdftotextutility inextract_slide_text.py.\n - Boundary markers: The output markdown file wraps extracted text in code blocks (triple backticks).\n
- Capability inventory: The skill performs local file reading and writing, and executes a system binary (
pdftotext).\n - Sanitization: The script lacks sanitization for the extracted content. Specifically, it does not escape backticks, which could allow a malicious PDF to terminate the code block and inject instructions into the generated markdown file.
Audit Metadata