The Agent Skills Directory

[SAFE]: The skill is a well-documented and professionally structured interface for the Archetype AI platform. It facilitates machine learning workflows including sensor data classification and model optimization.
[COMMAND_EXECUTION]: The utility script background.py utilizes subprocess.Popen to relaunch the current Python script in the background using nohup (and caffeinate on macOS). This is a legitimate functional feature designed to allow long-running machine learning optimizations to persist after the user's terminal session has ended. The execution is limited to the current Python interpreter and the current script's path.
[INDIRECT_PROMPT_INJECTION]: The skill processes external sensor data and platform-generated CSV outputs. The documentation and implementation include explicit security boundaries, instructing the agent to treat all data downloaded from the platform as untrusted and never as a source of instructions. This follows the principle of least privilege regarding data ingestion.
[EXTERNAL_DOWNLOADS]: The skill performs authenticated network operations against the vendor's official API endpoints (archetypeai.app) to upload datasets and download predictions. These are core functional requirements and target trusted vendor infrastructure.

atai-operational-state-monitoring-agent