# prompt-engineering

A collection of Chinese-language example projects for prompt engineering, LLM applications, RAG and agents, spanning older LangChain 0.1 code and newer LangChain and LangGraph 1.0 examples covering middleware, structured output and workflows.

This repository is a collection of case-study projects for large-model application development, covering prompt engineering, LLM applications, retrieval-augmented generation and agents, with folders named after their application area for learning and reference. Earlier examples use LangChain 0.1; from October 2026 the newer LangChain and LangGraph 1.0 examples cover agents, RAG, middleware, model calls, context engineering, structured output and workflow graphs.

Highlighted examples include a deep-search engine built with FastAPI, Playwright and a GLM model that runs multi-round iterative web search and crawling for research reports; an automatic prompt optimizer built on a Zhipu GLM model that iterates a prompt against labelled examples; a tool-retrieval service combining dense, sparse, keyword and hybrid search with a Chroma database and hypothetical-question generation, reported at 94 percent recall; a chat agent built with LangGraph, FastAPI and SQLite that understands images across multi-turn conversation with session management; and a self-evolving agent that extracts experience from past tasks offline and reuses it to improve success on new ones.

Website: https://github.com/5zjk5/prompt-engineering