Chia1104

The registry covers structural code queries via a knowledge graph, semantic code search, GitHub PR/issue/review operations through the gh CLI, document processing for PDF/DOCX/PPTX/XLSX files, ML model serving with vLLM and llama.cpp, and creative generation through ComfyUI, AudioCraft, and Segment Anything.

Software engineers working with TypeScript, React, oRPC, and Turborepo monorepos; ML practitioners benchmarking or serving models; researchers writing NeurIPS/ICML papers; and productivity-focused users managing email triage, RSS feeds, meeting action items, and weekly planning reviews.

Skills are folder-based with native frontmatter declaring name, description, and triggers. Many run cross-platform on Linux, macOS, and Windows. Some require prerequisite commands like blogwatcher-cli, openhue, or box; ComfyUI skills include a hardware check script and automated setup via pipx or uvx.

Most skills carry MIT licenses, including GitHub operations, RSS feeds, oRPC, and ML serving skills. However, the xlsx, pdf, pptx, and docx document skills are proprietary with complete terms in their LICENSE.txt files, and frontend-design has custom license terms.

ML skills declare explicit dependencies: vLLM serving needs vllm, torch, and transformers; llama.cpp needs llama-cpp-python; AudioCraft needs torch 2.0+ and transformers 4.30+; research paper writing needs semanticscholar, arxiv, scipy, numpy, matplotlib, and SciencePlots on Linux or macOS.