Epic Summary

Develop a comprehensive suite of specialized Claude Code skills to automate and streamline the complex package onboarding process for the AIPCC wheels builder system. This initiative addresses the significant complexity involved in integrating Python packages with hardware-specific build requirements, multi-variant support, and custom Fromager hook implementations.

These plugins will be stored in our AIPCC AI Helpers .

Business Value

• Reduce Onboarding Time: Automate complex analysis and configuration generation that currently requires deep expertise
• Improve Consistency: Standardize package onboarding approaches across different complexity levels
• Lower Barrier to Entry: Enable team members to onboard packages without deep Fromager expertise
• Reduce Errors: Automated analysis and configuration generation reduces manual configuration mistakes
• Scale Operations: Handle increasing package onboarding demands with consistent quality

Problem Statement

The AIPCC wheels builder currently requires extensive manual analysis and configuration for onboarding new Python packages, particularly those with:

• Complex build requirements across multiple hardware variants (CPU, CUDA, ROCm, Gaudi, TPU, Spyre)
• Custom source resolution for packages without PyPI source distributions
• Hardware-specific environment variables and build dependencies
• Multi-stage build processes with external source dependencies
• Variant-specific configurations and pre-built package delivery

This manual process is time-consuming, error-prone, and requires deep expertise in Fromager hooks, package build systems, and hardware-specific compilation requirements.