This repository is a meticulously curated knowledge base and toolkit for AI video prompt engineering. Its core is a structured data pipeline: raw knowledge from official vendor documentation and YouTube masterclasses is reverse-engineered into standardized methodologies (21+ SOPs) and a centralized prompt database (`prompts/data/all-prompts.json`). This data is then exposed through a cross-model benchmark (`prompts/data/cross-model-matrix.json`), zero-dependency web tools (Prompt Browser, Cross-Model Matrix), and executable Claude Code Skills (`skills/`). The architecture is data-centric. The primary value lies in the curated, structured data assets and the rigorous methodology used to create them. The application layer (web tools, skills) is a thin, well-crafted consumer of this data. The repository actively manages its primary risk—model version drift—through an automated monitoring system (`scripts/monitor_models.py`) that hashes external endpoints and files issues on change. While primarily a data_research project, it exhibits strong application characteristics through its interactive tools and AI skill integrations. It is a living knowledge base, actively maintained with a clear contribution workflow and a strong emphasis on data provenance and reproducibility. The `cross-model-matrix.json` is a standout asset, providing a controlled benchmark for comparing prompt styles across 11 major AI video models.
SGLang is a high-performance inference serving system for large language models (LLMs) and multimodal models (vision, video, audio, diffusion). It provides an OpenAI-compatible API, a Python frontend DSL for programmable inference, and supports a wide range of hardware backends (NVIDIA CUDA, AMD ROCm, Intel XPU, Ascend NPU, Apple MLX). The architecture is a distributed serving runtime (SRT) with a scheduler, model executor, custom CUDA/Triton kernels, and advanced features like speculative decoding (EAGLE, MTP, NGRAM), disaggregated prefill/decode, hierarchical KV cache (HiCache), and quantization (FP8, NVFP4, AWQ, GPTQ). It also includes a Rust-based model gateway for routing and a multimodal generation pipeline for diffusion models. The system is designed for production-scale deployment, powering trillions of tokens daily across hundreds of thousands of GPUs.