The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Python script (scripts/decision_tree_ev.py) to process user data. The execution is handled locally by the agent and does not involve risky shell commands or dynamic code generation.- [DATA_EXFILTRATION]: The skill operates entirely on provided configuration files. No network operations or unauthorized data access patterns were detected. Analysis of external references shows no connections to untrusted domains.- [PROMPT_INJECTION]: The instructions are clearly defined for strategic modeling. There are no attempts to override safety filters or bypass agent instructions.- [SAFE]: The Python script implements standard expected value and maximin calculations. It includes basic validation to ensure scenario probabilities sum to 1.0. No obfuscation, persistence mechanisms, or privilege escalation patterns were found.

war-gaming