The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to handle image conversion and environment setup. It utilizes the macOS sips tool to convert HEIC files to PNG and executes pip and brew commands to install necessary OCR and PDF libraries.
[REMOTE_CODE_EXECUTION]: The skill dynamically generates and executes a Python script at runtime. This script is used to center the original document image and format the English translation into a two-page PDF using the reportlab library.
[EXTERNAL_DOWNLOADS]: To perform its tasks, the skill installs several standard third-party libraries including pillow, reportlab, easyocr, and pytesseract. It also downloads the Tesseract OCR engine using Homebrew.
[PROMPT_INJECTION]: The skill has a potential indirect prompt injection surface because it processes text extracted from user-provided images via OCR.
Ingestion points: Text content is ingested from images provided by the user in SKILL.md.
Boundary markers: There are no specific instructions or delimiters provided to the agent to distinguish between the document content and potential instructions embedded in that content.
Capability inventory: The agent can execute shell commands (sips, pip, brew) and generate/execute Python scripts.
Sanitization: No sanitization or filtering is performed on the extracted text before it is translated or used in the PDF generation process.

visa-doc-translate