The Agent Skills Directory

[COMMAND_EXECUTION]: The serial_mux.py script uses subprocess.Popen to spawn a background multiplexing service using the system utility socat and its own Python process. The implementation uses a list of arguments rather than a raw shell string, which is a security best practice to prevent shell injection vulnerabilities. The purpose is to allow multiple applications to access the same hardware serial port simultaneously.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external serial devices, which represents an indirect prompt injection attack surface if the agent interprets embedded instructions in the stream.
Ingestion points: The serial data stream is read and processed in scripts/serial_monitor.py, scripts/serial_hex.py, and scripts/serial_log.py.
Boundary markers: The output is streamed as raw text or JSON Lines without specific boundary delimiters to isolate data from instructions.
Capability inventory: The skill has capabilities to write logs to the file system, send data back to the serial device, and manage a local TCP server for multiplexing.
Sanitization: Standard UTF-8 decoding with error replacement is used to sanitize incoming bytes for display.

serial