The Agent Skills Directory

[SAFE]: The skill implements secure credential handling by utilizing environment variable names rather than storing raw secrets in configuration files. It specifically employs sshpass -e to prevent password exposure in shell command history or process monitoring tools.
[COMMAND_EXECUTION]: The provided scripts/uart_session.py script uses subprocess calls for legitimate system operations, such as checking for device access locks using fuser and offering an interactive path to install necessary dependencies like pyserial from official repositories.
[PROMPT_INJECTION]: The UART transport implementation in uart_session.py effectively mitigates potential indirect prompt injection from the target device by using unique, randomly generated UUID sentinels for shell prompts and command exit codes, ensuring the agent correctly parses device output without being misled by its content.

jetson-validate-image