pdf-processing-openai

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • Privilege Escalation (HIGH): The skill instructions include sudo apt-get install, which grants the agent the ability to execute commands with root privileges and perform unauthorized system-level changes.
  • Command Execution (HIGH): The skill utilizes system commands (pdftoppm) with unquoted variables ($INPUT_PDF, $OUTPUT_PREFIX). If an attacker provides a PDF with a malicious filename (e.g., ; touch exploit;), it could lead to arbitrary command execution.
  • Indirect Prompt Injection (LOW):
  • Ingestion points: The skill is designed to process external PDF files via pdfplumber and pypdf which may contain malicious instructions.
  • Boundary markers: None are present in the prompt to separate data from instructions.
  • Capability inventory: The skill has access to shell execution, package managers, and file writing.
  • Sanitization: No evidence of input validation or sanitization for filenames or document content is provided.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:38 PM