Integrates local LLM inference into Go applications using the yzma library and llama.cpp.
Yzma is a specialized Claude Code skill designed for developers building Go applications with local AI capabilities. It streamlines the integration of llama.cpp, enabling seamless support for GGUF models across text generation, vision-language processing (VLM), embeddings, and tool-calling. This skill provides standardized implementation patterns for library loading, model management, and efficient resource handling, allowing you to leverage CPU or GPU acceleration via CUDA, Vulkan, or Metal directly within the Go ecosystem.
Key Features
01Local LLM inference via llama.cpp integration
02Standardized patterns for tool calling and vector embeddings
031 GitHub stars
04Hardware acceleration setup for CUDA, Vulkan, and Metal
05Automated resource management and memory cleanup workflows
06Support for GGUF format text and vision-language models
Use Cases
01Generating high-performance vector embeddings for local RAG systems
02Building privacy-focused offline chat applications in Go
03Implementing local vision-language processing for automated image analysis
