The Agent Skills Directory

[DATA_EXFILTRATION]: The skill uses a Swift-based component called events-log to capture keystrokes, mouse clicks, and cursor movements. This recording of system-wide input is a high-risk behavior that can lead to the exposure of sensitive data, including credentials and private communications.\n- [COMMAND_EXECUTION]: The functionality requires the user to grant 'Accessibility' and 'Input Monitoring' permissions to the terminal on macOS. These permissions allow the software to monitor all user interactions with the operating system, bypassing standard application isolation.\n- [COMMAND_EXECUTION]: The studio.py component launches a local web server to provide an editing interface, which creates an exposed network port on the local machine.\n- [DATA_EXFILTRATION]: The skill processes external event log data, creating a surface for indirect prompt injection:\n
Ingestion points: Reads and processes in.events.jsonl files in polish.py and render.py to drive video overlays.\n
Boundary markers: No delimiters or safety instructions are defined to separate data from instructions within the event logs.\n
Capability inventory: The skill utilizes ffmpeg for command execution, performs file system writes, and runs a local web server.\n
Sanitization: No evidence of input validation or sanitization for the event log content is provided.

screenstudio-alt