The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for unpacking asset archives using shell commands such as tar xzf. It also includes a Python script scripts/unpack.py which utilizes the tarfile library to programmatically extract experiment files to the filesystem.
[COMMAND_EXECUTION]: The skill relies on external binaries arduino-cli or arduino to facilitate the compilation and deployment of microcontroller firmware, which involves executing commands to interact with hardware.
[PROMPT_INJECTION]: The skill processes contents from local compressed archives, presenting a potential surface for indirect prompt injection if the source assets contain instructions meant to influence the agent's behavior.
Ingestion points: Local compressed archives in the assets/ directory, including .ino firmware files and .html simulators.
Boundary markers: The instructions do not define specific delimiters or "ignore" warnings when reading the content of extracted files.
Capability inventory: The skill has the capability to write files to the local disk and execute hardware-related command-line tools.
Sanitization: The scripts/unpack.py script performs extraction using tarfile.extractall() without validating file paths, which is a common security oversight for archive handling.

@6171/physics-lab-experiment-suite