The Agent Skills Directory

[SAFE]: The skill uses well-known and trusted Python libraries (onnx, onnxruntime) to perform model quantization tasks. The code patterns follow standard documentation for these libraries.
[SAFE]: No network operations, hardcoded credentials, or suspicious file system access patterns were identified. The file paths used (models/, data/calibration/) are local and consistent with the stated purpose of model optimization.
[SAFE]: The skill does not contain any obfuscated code, prompt injection attempts, or unauthorized command execution. It focuses entirely on mathematical transformations of machine learning models and metric evaluation.

fp16_int8_quantization