The Agent Skills Directory

[SAFE]: The skill logic is implemented in pure Python using only standard library modules (math, random, json). It contains no hardcoded credentials, network exfiltration patterns, or persistence mechanisms.
[COMMAND_EXECUTION]: The testing script scripts/test_ab_stats.py utilizes subprocess.run to execute the primary ab_stats.py script against known fixtures. This execution is confined to the local environment and uses the system's current Python interpreter to call an internal script, representing standard and secure software testing practices.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external experiment data via JSON files or standard input. It mitigates potential data-driven attacks by implementing strict validation logic in scripts/ab_stats.py, including schema verification in validate_input and value-type checking in _validate_binary to ensure input consistency before performing calculations.

experiment-stats