Fourier has made a groundbreaking move by officially open-sourcing the extensive full-size humanoid robot dataset known as Fourier ActionNet, alongside the world's first comprehensive full-process toolchain. The initial release boasts over 30,000 high-quality entries, comprising real-machine training data that spans a diverse array of dexterous hand training scenarios with multiple degrees of freedom and hand task imitation learning. This comprehensive dataset is now accessible to developers and scientific research institutions globally, offering a seamless solution encompassing data collection through to deployment. Fourier ActionNet encompasses all model task training data from Fourier's GRx series, leveraging a visual language model for automated annotation and manual verification to guarantee data precision. The introduction of this full-process toolchain significantly lowers the barriers to entry for research and development in humanoid robot technology.