The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to dynamically generate and execute a Python script to create the final PDF translation. This script incorporates text content extracted via OCR from untrusted user-provided images, creating a path for code injection if the agent does not properly sanitize the content before placing it into the script code.
[COMMAND_EXECUTION]: The agent is directed to use system-level command line tools such as sips for image conversion and brew for package management.
[PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection because it processes text extracted from images via OCR without any boundary markers, sanitization, or explicit instructions to the agent to ignore commands found within the document text.
[PROMPT_INJECTION]: The skill explicitly commands the agent to perform its entire workflow automatically without asking for user confirmation, which bypasses human-in-the-loop security oversight for potentially dangerous operations like script execution.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of multiple external dependencies from public registries, including Python packages like easyocr, pytesseract, pillow, and reportlab, as well as system binaries like tesseract.

visa-doc-translate