Prompt Engineering Guide for Large Models
A Chinese-language book on prompt engineering, covering prompt structure, chain-of-thought and multimodal prompting through to enterprise PromptOps, agent workflows and retrieval-augmented generation.
This is a book-length guide to prompt engineering for large language models, written in Chinese and organized from introductory concepts to advanced application. It covers the core elements of a prompt, such as role, background, task and constraints, advanced techniques such as few-shot learning, chain-of-thought and ReAct, prompting for multimodal input across text, image, audio and video, and building prompt chains and agent workflows.
Later sections cover prompt optimization for retrieval-augmented generation, designing secure enterprise-grade prompt architectures, building a workflow to test, evaluate and continuously improve prompts, and automated prompt generation and optimization. It targets AI application developers, product managers and designers, content creators, data scientists and machine learning engineers, and enterprise decision-makers, and recommends a production-readiness sequence of prompt contracts, evaluation, context, deterministic workflows, single-agent and finally multi-agent systems.
It can be read online through GitBook or downloaded as a PDF from the project's GitHub Releases.