gonzalescataphatic400/Qwen3.5-TurboQuant-MLX-LM
Accelerate Qwen3.5 KVCache with TurboQuant MLX-LM tools for prompt caching, quantized attention, and research-ready evals
Accelerate Qwen3.5 KVCache with TurboQuant MLX-LM tools for prompt caching, quantized attention, and research-ready evals
npx skills add gonzalescataphatic400/Qwen3.5-TurboQuant-MLX-LMAccelerate Qwen3.5 KVCache with TurboQuant MLX-LM tools for prompt caching, quantized attention, and research-ready evals
AI agent methodology skill for Codex, Claude Code, and OpenCode: dialectical analysis, software architecture reasoning, product strategy, and project retrospection.唯物辩证批判思维 Skill — 基于毛选方法论的 AI 底层思维框架的 coding 框架
AI coding chat portrait · 分析你跟 Claude Code / OpenCode / Kiro / Trae 的聊天记录,生成可视化性格画像。Self-reflection meta skill with conversation analytics, bilingual (zh/en), 100% local.
Automate legal research, citation verification, and DOCX document processing with reusable AI agent skills compatible with local agent runtimes.
Deploy a local, multi-user RAG system to query PDF and DOCX documents using a local LLM without cloud or API dependencies.
Football data analytics toolkit for Claude Code. Covers Opta, StatsBomb, Wyscout, SportMonks, and free sources.
Diagnose and fix retention problems using behavior design (B=MAP). Use when the user mentions "users drop off", "activation rate", "onboarding friction", "retention metrics", "why users dont complete", "churn analysis", "user activation", or "aha moment". Also trigger when analyzing cohort retention curves, designing activation milestones, reducing time-to-value for new users, or investigating why users stop after their first session. Covers the Ability Chain, prompt design, and tiny behaviors that compound. For habit loops and variable rewards, see hooked-ux. For intrinsic motivation, see drive-motivation.