Agent FrameworkAgent 框架、多智能体系统、LLM 工程实践。https://leejuju.me/ECOM1: E-commerce Agent Challenge 解读https://leejuju.me/blog/challenge/ecom1-agent-challenge/ecom1-agent-challenge/https://leejuju.me/blog/challenge/ecom1-agent-challenge/ecom1-agent-challenge/BitGN 电商智能体基准：覆盖购物、结账、退款、风控等业务域，评估 Agent 在真实业务约束下的工具调用和风险控制能力。Thu, 28 May 2026 00:00:00 GMT1. langchain-core 源码https://leejuju.me/blog/framework/langchain/1-langchain-core/readme/https://leejuju.me/blog/framework/langchain/1-langchain-core/readme/langchain-core 是 LangChain 的核心包。(之前)Sat, 16 May 2026 00:00:00 GMTlangchain2. langchain 源码https://leejuju.me/blog/framework/langchain/2-langchain/agent/readme/https://leejuju.me/blog/framework/langchain/2-langchain/agent/readme/10/07/2025 langchain 大改版后的代码吗，包含 agent 创建的工厂方法，有以下参数。特别重要的一点是，从 2025 年大改后，agent 底层走的是 graph 那一套。Tue, 16 Jun 2026 00:00:00 GMTlangchain3. langgraph 源码https://leejuju.me/blog/framework/langchain/3-langgraph/readme/https://leejuju.me/blog/framework/langchain/3-langgraph/readme/langgraph langchain 对于其 agent 执行逻辑的思考和重构，我倒是很好奇Tue, 16 Jun 2026 00:00:00 GMTlangchainAgent 记忆系统调研：Mem0、Zep、EverMemOShttps://leejuju.me/blog/memory/agent-memory-systems/https://leejuju.me/blog/memory/agent-memory-systems/调研多种 AI Agent 记忆系统方案，包括 Mem0、Zep 和 EverMemOS，分析它们在持久化记忆、混合检索和多 Agent 协等方面的架构设计与实践路径Mon, 22 Dec 2025 00:00:00 GMTAnthropic Building Effective Agentshttps://leejuju.me/blog/paper--engineering/building-effective-agents/building-effective-agents/https://leejuju.me/blog/paper--engineering/building-effective-agents/building-effective-agents/解读 Anthropic 多 Agent 研究系统的设计原则：Orchestrator 模式、并行检索、任务分解、工具设计等 7 个关键实践。Wed, 13 May 2026 00:00:00 GMT论文笔记：Is Graph All You Need?https://leejuju.me/blog/paper--engineering/is-graph-all-you-need/is-graph-all-you-need/https://leejuju.me/blog/paper--engineering/is-graph-all-you-need/is-graph-all-you-need/基于图的 IR 检索 vs Grep vs 向量 RAG，在 Agent Search 场景下的效果对比，以及 Chronos 框架的设计思路。Sun, 31 May 2026 00:00:00 GMTRAG 策略对比：Basic、Graph、Grep、Structuralhttps://leejuju.me/blog/rag/rag-challenge-2--3/rag-strategies-comparison/https://leejuju.me/blog/rag/rag-challenge-2--3/rag-strategies-comparison/深入对比四种 RAG 策略——基础 HardChunk、Graph RAG、Grep RAG 与 Structural RAG，基于 Enterprise RAG Challenge 2 数据集，从检索质量、语义理解和工程代价等维度进行系统评估Mon, 22 Dec 2025 00:00:00 GMT