Searches and extracts API details from NVIDIA CUDA-X libraries to streamline GPU-accelerated development.
This skill enables Claude to browse and search comprehensive NVIDIA CUDA-X library documentation, including cuBLAS, cuTENSOR, cuDNN, and more. It utilizes specialized tools to query Sphinx inventory files, find specific API symbols, and extract full function signatures and documentation trees directly into the chat interface. It is essential for developers working on high-performance computing, deep learning, or GPU-accelerated math who need quick, accurate access to library specifications without manual web searching.
Key Features
01Support for the full suite of CUDA-X libraries like cuQuantum and cuSOLVER
02Cross-library API discovery via Sphinx inventory mapping
030 GitHub stars
04Detailed documentation extraction including function signatures and parameters
05Domain-specific filtering for C, C++, and Python bindings
06Fuzzy search support for finding specific CUDA operations and types
Use Cases
01Searching for specific tensor decomposition functions in the cuQuantum library
02Verifying API availability and obtaining direct documentation URLs for CUDA math functions
03Retrieving C++ function signatures and parameter descriptions for cuBLAS GEMM operations
