从 AI 认知到 Agent 工程，从财税合规到产品方法论 -- 452+ 篇精选文章，系统化构建 AI 能力https://lingque.academy/zh-cnhttps://lingque.academy/learn/ai-literacy/%E4%BC%81%E4%B8%9Aai%E5%BC%80%E5%8F%91%E5%B7%A5%E4%BD%9C%E6%B5%81%E8%90%BD%E5%9C%B0%E8%93%9D%E5%9B%BE/https://lingque.academy/learn/ai-literacy/%E4%BC%81%E4%B8%9Aai%E5%BC%80%E5%8F%91%E5%B7%A5%E4%BD%9C%E6%B5%81%E8%90%BD%E5%9C%B0%E8%93%9D%E5%9B%BE/企业级 AI 工作流 · 落地版 每个岗位如何把 AI 开发嵌入工作流 面向大中型研发组织的执行蓝图。核心不是“人人都去用 AI”，而是把岗位动作拆解为 信息处理、判断决策、执行产出 三层，先让 AI 吃掉高频、低风险、可验证、可模板化的环节， 再逐步进入复杂协同链路，最终沉淀为组织级平台能力。 战略原则 先嵌入，后重构 推进单位 岗位动作 核心路径 模板 + 门禁 + 度量 最终目标 组织级...Wed, 08 Apr 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/%E7%A0%94%E5%8F%91%E9%BB%91%E8%AF%9D_cheatsheet/https://lingque.academy/learn/ai-literacy/%E7%A0%94%E5%8F%91%E9%BB%91%E8%AF%9D_cheatsheet/研发黑话 Cheatsheet 给产品、研发、AI 应用、Agent、平台架构场景用的高频术语速查表。目标不是背词，而是知道每个词在团队里通常暗示什么。 研发流程 AI / Agent 架构与平台 数据与评估 工程效率 上线运维 01 基础研发黑话 02 AI / Agent 黑话 03 数据 / 检索 / RAG 04 架构 / 平台 / 系统 05 质量 / 测试 / 稳定性 06 项目管理...Fri, 03 Apr 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/notes/https://lingque.academy/learn/ai-literacy/notes/Notes: OpenClaw培训-S1-1.html UI/UX Optimization Previous turn applied text-wrap: pretty for CJK typography. Added smooth hover transitions to cards. Fixed print styles. Need to verify button hierarchy...Wed, 01 Apr 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/robot-speech-linda/https://lingque.academy/learn/ai-literacy/robot-speech-linda/WHAT'S HAPPENED? A little robot just droppeda secret note... It says: "Come learn about us!" Let's explore the world of ROBOTS! WHAT IS A ROBOT? Smart Machines That Move A robot is a smart machine It...Sat, 28 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/shenzhen-tech-speech-linda/https://lingque.academy/learn/ai-literacy/shenzhen-tech-speech-linda/ENGLISH SPEECH CONTEST 2026 ShenzhenThe City of Tomorrow Where Innovation Meets the Future Linda Wen (文婉甄) | Grade 3 | Qianhai Gangwan School | 前海港湾小学 GUESS WHAT? Have you ever seen a carwith NO...Sat, 28 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/ai_programming_engineering_methodologies_research_note_2026/https://lingque.academy/learn/ai-literacy/ai_programming_engineering_methodologies_research_note_2026/研究笔记 AI 编程 工程方法论 2026 版 AI 编程工程方法论研究笔记 这份文档把 2025–2026 年最有代表性的 AI-native 软件工程方法，压缩成一个可操作的研究框架：从 Spec、Context、Tool、Agent Workflow、Eval 到 Harness，最后收束到 Compound Engineering 的复利系统观。...Wed, 25 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/strategy-research/%E7%BC%96%E7%A8%8B%E8%BF%9B%E5%8C%96%E5%8F%B2%E6%8A%A5%E5%91%8A/https://lingque.academy/learn/strategy-research/%E7%BC%96%E7%A8%8B%E8%BF%9B%E5%8C%96%E5%8F%B2%E6%8A%A5%E5%91%8A/Programming Evolution Report 编程进化史 从手工编程 → Tab 编程 → Vibe 编程 → 工程编程 → Harness 编程。这不是工具升级史，而是软件生产控制权从“写代码”迁移到“定义约束、组织执行、设计验收”的演化史。 一、总览：五个阶段不是替代关系，而是控制权重心迁移 阶段 核心动作 人的主要角色 机器的主要角色 稀缺能力 手工编程 逐行写 编码器 执行器...Mon, 23 Mar 2026 00:00:00 GMTstrategy-researchhttps://lingque.academy/learn/ai-literacy/openclaw_lobster_fever_report/https://lingque.academy/learn/ai-literacy/openclaw_lobster_fever_report/Deep Thinking Report “龙虾热”背后的结构性成因：一份关于 OpenClaw 的深度思考报告 这不是一份产品测评，而是一份结构判断：OpenClaw 引发的“龙虾热”，本质上是 Agent 被市场第一次大规模重估为“数字劳动力”和“新生产资料”的社会事件。 主题：OpenClaw / Agent / 新入口 时间锚点：2026 年 3 月 方法：现象拆解 + 产业结构推演...Wed, 18 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/openclaw%E5%9F%B9%E8%AE%AD-s1-1-backup-2326/https://lingque.academy/learn/ai-literacy/openclaw%E5%9F%B9%E8%AE%AD-s1-1-backup-2326/INTERNAL TRAINING S2-1 OpenClaw从顾问到员工 财税多智能体平台 -- 内部技术培训 2026 Q1 | 产品经理 x 研发团队 | 基于蓝皮书 v1.0 + 社区实战 扫码签到 Sources: OpenClaw GitHub 318K+ Stars; ClawHub 13,700+ Published Skills; 杨彧鑫AI OpenClaw 蓝皮书...Tue, 17 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/task_plan/https://lingque.academy/learn/ai-literacy/task_plan/Task Plan: OpenClaw培训-S1-1.html UI/UX Optimization Objective Apply frontend UI/UX optimization SOP to OpenClaw培训-S1-1.html, ensuring high-quality composition, hierarchy, emphasis, and rhythm. Steps...Tue, 17 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/claude_code_official_tools_explained/https://lingque.academy/learn/ai-literacy/claude_code_official_tools_explained/Claude Code 官方工具全解 基于 Anthropic 官方文档整理，日期：2026-03-15。 先纠偏：“Claude Code 官方有 20 个工具”这个说法，已经不是当前官方文档口径。Anthropic 现在的 Tools reference 页面列出了 30...Sun, 15 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/strategy-research/%E5%86%85%E5%8F%82%E6%8A%A5%E5%91%8A4%E6%9C%9F-%E6%99%BA%E8%83%BD%E4%BD%93%E5%AF%92%E6%AD%A6%E7%BA%AA%E5%A4%A7%E7%88%86%E5%8F%91/https://lingque.academy/learn/strategy-research/%E5%86%85%E5%8F%82%E6%8A%A5%E5%91%8A4%E6%9C%9F-%E6%99%BA%E8%83%BD%E4%BD%93%E5%AF%92%E6%AD%A6%E7%BA%AA%E5%A4%A7%E7%88%86%E5%8F%91/内参系列 · Special Report · 第 IV 期 深度洞察 2026 年 3 月 8 日 · 产业战略 · 内参报告第 IV 期 AI 产业 · 战略指引 智能体寒武纪：五条产业脉动、三个中国机会与一场迫在眉睫的大洗牌 Product Hunt 日榜 65% 被 Agent 产品占据——这不是一个品类的崛起，而是整个软件产业底层逻辑的重写。本期从全球 AI...Sun, 08 Mar 2026 00:00:00 GMTstrategy-researchhttps://lingque.academy/learn/graph-rag/awesome-graphrag_index/https://lingque.academy/learn/graph-rag/awesome-graphrag_index/Awesome-GraphRAG (GraphRAG Survey) This repository contains a curated list of resources on graph-based retrieval-augmented generation (GraphRAG), which are classified according to "A Survey of Graph...Fri, 06 Mar 2026 00:00:00 GMTgraph-raghttps://lingque.academy/learn/model-engineering/deepeval_guide/https://lingque.academy/learn/model-engineering/deepeval_guide/The LLM Evaluation Framework Documentation | Metrics and Features | Getting Started | Integrations | DeepEval Platform DeepEval is a simple-to-use, open-source LLM evaluation framework, for...Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/model-engineering/lm_evaluation_harness/https://lingque.academy/learn/model-engineering/lm_evaluation_harness/Language Model Evaluation Harness Latest News [2025/12] CLI refactored with subcommands (run, ls, validate) and YAML config file support via --config. See the CLI Reference and Configuration Guide....Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/graph-rag/lightrag_guide/https://lingque.academy/learn/graph-rag/lightrag_guide/LightRAG: Simple and Fast Retrieval-Augmented Generation News [2025.11][New Feature]: Integrated RAGAS for Evaluation and Langfuse for Tracing. Updated the API to return retrieved contexts alongside...Fri, 06 Mar 2026 00:00:00 GMTgraph-raghttps://lingque.academy/learn/model-engineering/llamafactory_guide/https://lingque.academy/learn/model-engineering/llamafactory_guide/Used by Amazon, NVIDIA, Aliyun, etc. Supporters Warp, the agentic terminal for developersAvailable for MacOS, Linux, & Windows Easily fine-tune 100+ large language models with zero-code CLI and Web...Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/graph-rag/neo4j_llm_graph_builder/https://lingque.academy/learn/graph-rag/neo4j_llm_graph_builder/Knowledge Graph Builder Transform unstructured data (PDFs, DOCs, TXTs, YouTube videos, web pages, etc.) into a structured Knowledge Graph stored in Neo4j using the power of Large Language Models...Fri, 06 Mar 2026 00:00:00 GMTgraph-raghttps://lingque.academy/learn/model-engineering/trl_alignment_guide/https://lingque.academy/learn/model-engineering/trl_alignment_guide/TRL - Transformer Reinforcement Learning A comprehensive library to post-train foundation models What's New OpenEnv Integration: TRL now supports OpenEnv, the open-source framework from Meta for...Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/model-engineering/unsloth_guide/https://lingque.academy/learn/model-engineering/unsloth_guide/Train gpt-oss, DeepSeek, Gemma, Qwen & Llama 2x faster with 70% less VRAM! Train for Free Notebooks are beginner friendly. Read our guide. Add dataset, run, then deploy your trained model. Model Free...Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/graph-rag/fast-graphrag_guide/https://lingque.academy/learn/graph-rag/fast-graphrag_guide/Streamlined and promptable Fast GraphRAG framework designed for interpretable, high-precision, agent-driven retrieval workflows. Install | Quickstart | Community | Report Bug | Request Feature...Fri, 06 Mar 2026 00:00:00 GMTgraph-raghttps://lingque.academy/learn/model-engineering/vllm_guide/https://lingque.academy/learn/model-engineering/vllm_guide/Easy, fast, and cheap LLM serving for everyone | Documentation | Blog | Paper | Twitter/X | User Forum | Developer Slack | We have built a vllm website to help you get started with vllm. Please visit...Fri, 06 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/strategy-research/ai_%E8%81%8C%E4%B8%9A%E6%9A%B4%E9%9C%B2%E5%9B%BE_%E6%B7%B1%E5%BA%A6%E8%A7%A3%E8%AF%BB%E6%8A%A5%E5%91%8A/https://lingque.academy/learn/strategy-research/ai_%E8%81%8C%E4%B8%9A%E6%9A%B4%E9%9C%B2%E5%9B%BE_%E6%B7%B1%E5%BA%A6%E8%A7%A3%E8%AF%BB%E6%8A%A5%E5%91%8A/AI & Labour Market March 2026 Deep Analysis "理论能做"与"现实已用"之间的裂缝 基于 Anthropic 2026 年 3 月研究及用户提供图表线索，对 AI 在不同职业类别中的潜在覆盖与真实落地差距进行系统解读...Fri, 06 Mar 2026 00:00:00 GMTstrategy-researchhttps://lingque.academy/learn/ai-literacy/gemma-%E5%9C%A8%E4%B8%AD%E5%9B%BD-to-b-%E5%9C%BA%E6%99%AF%E4%B8%AD%E7%9A%84%E4%BB%B7%E5%80%BC-%E8%BE%B9%E7%95%8C%E4%B8%8E%E6%9B%BF%E4%BB%A3%E8%B7%AF%E7%BA%BF/https://lingque.academy/learn/ai-literacy/gemma-%E5%9C%A8%E4%B8%AD%E5%9B%BD-to-b-%E5%9C%BA%E6%99%AF%E4%B8%AD%E7%9A%84%E4%BB%B7%E5%80%BC-%E8%BE%B9%E7%95%8C%E4%B8%8E%E6%9B%BF%E4%BB%A3%E8%B7%AF%E7%BA%BF/Briefing March 6th 2026 China Technology Gemma 在中国 To-B 场景中的价值、边界与替代路线 这不是一份"模型安利"，而是一份更接近尽调的报告：它同时问技术、问法务、问业务， 问的不是"它聪不聪明"，而是"它能不能真的被企业接住"。 真正的企业选型，常常不是在寻找最会背题的模型，而是在寻找那个更愿意被预算、流程、审计、...Fri, 06 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/graph-rag/%E7%81%B5%E9%98%99%E9%AB%98%E7%BB%B4%E7%9F%A5%E8%AF%86%E5%9B%BE%E8%B0%B1-agent-%E5%B9%B3%E5%8F%B0%E6%9E%B6%E6%9E%84/https://lingque.academy/learn/graph-rag/%E7%81%B5%E9%98%99%E9%AB%98%E7%BB%B4%E7%9F%A5%E8%AF%86%E5%9B%BE%E8%B0%B1-agent-%E5%B9%B3%E5%8F%B0%E6%9E%B6%E6%9E%84/Architecture Overview 灵阙高维知识图谱Agent 平台架构 基于多 Agent 协同、高维知识图谱与大模型推理引擎构建的业务财务税务合规数智人平台，面向复杂监管环境提供端到端的感知、推理与执行能力。 业财税合规 多 Agent 协同 知识图谱 时序推理 数智人平台 Section I 核心架构层级 Core Architecture Layers — Top → Down...Wed, 04 Mar 2026 00:00:00 GMTgraph-raghttps://lingque.academy/learn/product-design/ai%E7%BC%96%E7%A8%8B%E6%8A%A5%E5%91%8A-gemini/https://lingque.academy/learn/product-design/ai%E7%BC%96%E7%A8%8B%E6%8A%A5%E5%91%8A-gemini/AI编程工具实战报告（基于Gemini） 目录 第一部分：新的开发范式：从AI助手到AI队友 1.1 “模板代码”的终结：AI如何重塑软件开发生命周期 1.2 理解AI辅助的光谱：一个四层模型 1.3 Agentic AI登场：从“编写代码”到“解决问题”的转变 第二部分：描绘全球AI编码工具版图 2.1 类别深度剖析：选择你的“副驾驶” 2.2 焦点：科技巨头的“生态系统之战”...Tue, 03 Mar 2026 00:00:00 GMTproduct-designhttps://lingque.academy/learn/ai-literacy/openclaw-%E4%B8%BB%E5%8A%A8%E6%99%BA%E8%83%BD%E4%BD%93%E7%AB%9E%E8%B5%9B-%E7%A1%85%E8%B0%B7%E4%B8%8E%E4%B8%AD%E5%9B%BD%E5%A4%A7%E5%8E%82%E8%B7%AF%E7%BA%BF%E5%AF%B9%E6%AF%94-%E6%B7%B1%E5%BA%A6%E7%A0%94%E7%A9%B6/https://lingque.academy/learn/ai-literacy/openclaw-%E4%B8%BB%E5%8A%A8%E6%99%BA%E8%83%BD%E4%BD%93%E7%AB%9E%E8%B5%9B-%E7%A1%85%E8%B0%B7%E4%B8%8E%E4%B8%AD%E5%9B%BD%E5%A4%A7%E5%8E%82%E8%B7%AF%E7%BA%BF%E5%AF%B9%E6%AF%94-%E6%B7%B1%E5%BA%A6%E7%A0%94%E7%A9%B6/OpenClaw 主动智能体竞赛：硅谷与中国大厂路线对比 从“更聪明的模型”到“更可靠的执行系统”：同一场竞赛，两套产业答案。 深度研究 经济学人风格 更新时间：2026-03-03 编号：OC-AGENT-260303 摘要 核心判断：OpenClaw...Tue, 03 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/strategy-research/%E5%86%85%E5%8F%82%E6%8A%A5%E5%91%8A3%E6%9C%9F-%E9%BE%99%E8%99%BE%E5%82%AC%E7%94%9F%E7%9A%84%E4%B8%A4%E4%B8%AA%E4%B8%96%E7%95%8C/https://lingque.academy/learn/strategy-research/%E5%86%85%E5%8F%82%E6%8A%A5%E5%91%8A3%E6%9C%9F-%E9%BE%99%E8%99%BE%E5%82%AC%E7%94%9F%E7%9A%84%E4%B8%A4%E4%B8%AA%E4%B8%96%E7%95%8C/内参系列 · Special Report · 第 III 期 深度洞察 2026 年 3 月 3 日 · 技术与产业 · 内参报告第 III 期 主动智能体 · 全球竞赛 龙虾催生的两个世界：从 pi 引擎到 Frontier，OpenClaw 智能体竞赛的全球路径分裂与终局推演 当硅谷把智能体当"新员工"来管理，中国大厂把它变成"云上即插即用的 SKU"，而国产模型厂商纷纷推出...Tue, 03 Mar 2026 00:00:00 GMTstrategy-researchhttps://lingque.academy/learn/agent-engineering/ai_agents_for_beginners/https://lingque.academy/learn/agent-engineering/ai_agents_for_beginners/AI Agents for Beginners 15 Lessons -- Get Started Building AI Agents. A structured course by Microsoft covering AI agent fundamentals, agentic frameworks, design patterns, tool use, multi-agent...Mon, 02 Mar 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/ai-literacy/anthropic_courses_overview/https://lingque.academy/learn/ai-literacy/anthropic_courses_overview/Anthropic Official Courses Official educational courses from Anthropic covering Claude API fundamentals, prompt engineering, real-world prompting patterns, tool use, and prompt evaluations. Each...Mon, 02 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/applied-ai/awesome_llm_apps/https://lingque.academy/learn/applied-ai/awesome_llm_apps/Awesome LLM Apps A curated collection of Awesome LLM apps built with RAG, AI Agents, Multi-agent Teams, MCP, Voice Agents, and more. This repository features LLM apps that use models from OpenAI ,...Mon, 02 Mar 2026 00:00:00 GMTapplied-aihttps://lingque.academy/learn/ai-literacy/brex_prompt_engineering_guide/https://lingque.academy/learn/ai-literacy/brex_prompt_engineering_guide/Brex's Prompt Engineering Guide This guide was created by Brex for internal purposes. It's based on lessons learned from researching and creating Large Language Model (LLM) prompts for production use...Mon, 02 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/ai-literacy/generative_ai_for_beginners/https://lingque.academy/learn/ai-literacy/generative_ai_for_beginners/Generative AI for Beginners 21 Lessons -- Get Started Building with Generative AI. A comprehensive course by Microsoft Cloud Advocates covering everything from GenAI fundamentals to building...Mon, 02 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/model-engineering/huggingface_smol_course/https://lingque.academy/learn/model-engineering/huggingface_smol_course/HuggingFace Smol Course A practical, hands-on course on aligning language models. Covers instruction tuning (SFT), evaluation, preference alignment (DPO), and vision-language models. Everything runs...Mon, 02 Mar 2026 00:00:00 GMTmodel-engineeringhttps://lingque.academy/learn/ai-literacy/llm_course_roadmap/https://lingque.academy/learn/ai-literacy/llm_course_roadmap/The LLM course is divided into three parts: LLM Fundamentals is optional and covers fundamental knowledge about mathematics, Python, and neural networks. The LLM Scientist focuses on building the...Mon, 02 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/architecture/openai_cookbook_curated/https://lingque.academy/learn/architecture/openai_cookbook_curated/OpenAI Cookbook -- Curated Best Practices A curated selection of the most impactful articles from OpenAI's official Cookbook (72K+ stars). Covers LLM fundamentals, agent building, embeddings, vector...Mon, 02 Mar 2026 00:00:00 GMTarchitecturehttps://lingque.academy/learn/ai-literacy/prompt_engineering_guide/https://lingque.academy/learn/ai-literacy/prompt_engineering_guide/Prompt Engineering Guide Prompt engineering is a relatively new discipline for developing and optimizing prompts to efficiently use language models (LMs) for a wide variety of applications and...Mon, 02 Mar 2026 00:00:00 GMTai-literacyhttps://lingque.academy/learn/agent-engineering/ai-agent%E7%9A%84%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E4%BC%98%E5%8C%96%E7%AD%96%E7%95%A5/https://lingque.academy/learn/agent-engineering/ai-agent%E7%9A%84%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E4%BC%98%E5%8C%96%E7%AD%96%E7%95%A5/AI Agent 的工具调用优化策略 概述 工具调用是 Agent 与外部世界交互的基础能力。但在生产环境中，朴素的工具调用面临多个问题：延迟高（每次调用都是网络请求）、成本高（每次调用消耗 token）、不稳定（API 可能失败）、安全风险（Agent 可能调用不该调用的工具）。 本文从并行调用、缓存策略、错误恢复、成本控制和工具选择优化五个维度，给出工程级的解决方案。 一、并行工具调用 问题...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agentic-coding-assistant%E6%9E%B6%E6%9E%84%E8%A7%A3%E6%9E%90/https://lingque.academy/learn/agent-engineering/agentic-coding-assistant%E6%9E%B6%E6%9E%84%E8%A7%A3%E6%9E%90/Agentic Coding Assistant 架构解析 代码生成 Agent 的上下文工程、编辑应用策略、测试驱动循环与自主编程范式 引言 2024-2025 年，Coding Assistant 从"自动补全"进化到"自主编程"。Cursor、GitHub Copilot Workspace、Devin、Claude...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E4%B8%8E%E4%BA%BA%E7%B1%BB%E5%8D%8F%E4%BD%9C%E7%9A%84%E4%BA%A4%E4%BA%92%E8%AE%BE%E8%AE%A1/https://lingque.academy/learn/agent-engineering/agent%E4%B8%8E%E4%BA%BA%E7%B1%BB%E5%8D%8F%E4%BD%9C%E7%9A%84%E4%BA%A4%E4%BA%92%E8%AE%BE%E8%AE%A1/Agent 与人类协作的交互设计 人机协作模式、审批流程与升级协议的工程化实践 人机协作的设计原则 Agent 不是取代人类，而是与人类协作。关键设计原则： 透明性（Transparency）：Agent 必须让人类理解它在做什么、为什么这么做 可控性（Controllability）：人类随时可以介入、修改、中断 Agent 的行为 适度自主（Calibrated...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E5%AE%89%E5%85%A8%E4%B8%8E%E6%8A%A4%E6%A0%8F%E8%AE%BE%E8%AE%A1/https://lingque.academy/learn/agent-engineering/agent%E5%AE%89%E5%85%A8%E4%B8%8E%E6%8A%A4%E6%A0%8F%E8%AE%BE%E8%AE%A1/Agent 安全与护栏设计 Prompt Injection 防御、输出验证、沙盒隔离、权限系统与内容过滤实战 引言 当 Agent 具备了工具调用、网络访问和代码执行能力后，安全问题不再是理论威胁，而是实际的攻击面。一次成功的 Prompt Injection 可以让 Agent 泄露系统提示、调用未授权工具、甚至执行恶意代码。更危险的是，Agent...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E5%AE%89%E5%85%A8%E4%B8%8E%E6%9D%83%E9%99%90%E6%8E%A7%E5%88%B6%E6%A1%86%E6%9E%B6/https://lingque.academy/learn/agent-engineering/agent%E5%AE%89%E5%85%A8%E4%B8%8E%E6%9D%83%E9%99%90%E6%8E%A7%E5%88%B6%E6%A1%86%E6%9E%B6/Agent 安全与权限控制框架 构建安全可控的 AI Agent 系统：从沙盒隔离到权限模型 Agent 安全的本质挑战 Agent 与传统软件的根本区别：Agent 的行为是非确定性的。 同一个 Prompt，不同的上下文，可能产生完全不同的工具调用序列。这意味着传统的白名单/黑名单安全模型无法完全覆盖 Agent 的行为空间。 传统软件安全模型： 输入 ──→ 确定性逻辑 ──→...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E4%BC%98%E5%8C%96%E7%AD%96%E7%95%A5/https://lingque.academy/learn/agent-engineering/agent%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E4%BC%98%E5%8C%96%E7%AD%96%E7%95%A5/Agent 工具调用优化策略 从工具选择到并行执行，构建高效可靠的 Agent 工具调用体系 工具调用的核心问题 Agent 的能力边界由其可用的工具集决定。工具调用的效率和可靠性直接决定了 Agent 的实际表现。核心挑战包括： 选择问题：面对数十甚至数百个工具，如何让 Agent 准确选择最合适的工具 参数问题：工具参数的构造错误是 Agent 失败的首要原因...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E6%A8%A1%E5%BC%8F%E4%B8%8Efunction-calling/https://lingque.academy/learn/agent-engineering/agent%E5%B7%A5%E5%85%B7%E8%B0%83%E7%94%A8%E6%A8%A1%E5%BC%8F%E4%B8%8Efunction-calling/Agent 工具调用模式与 Function Calling 工具 Schema 设计、并行工具调用、结构化输出、错误恢复与工具选择策略 引言 工具调用（Function Calling / Tool Use）是 Agent 从"只能说"到"能做事"的关键跳跃。LLM 本身只能生成文本，但通过工具调用，它可以查询数据库、调用 API、操作文件系统——将思考转化为行动。...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E7%BC%96%E6%8E%92%E5%BC%95%E6%93%8E-langgraph-vs-crewai-vs-autogen/https://lingque.academy/learn/agent-engineering/agent%E7%BC%96%E6%8E%92%E5%BC%95%E6%93%8E-langgraph-vs-crewai-vs-autogen/Agent 编排引擎：LangGraph vs CrewAI vs AutoGen 三大主流 Agent 编排框架的架构对比与选型指南 编排引擎解决什么问题 Agent 编排引擎负责管理 Agent 的生命周期、工作流控制、状态管理和工具调度。核心要回答的问题： 流程控制：Agent 的执行路径如何定义？是固定的还是动态的？ 状态管理：多步骤执行中的中间状态如何持久化和传递？ 多 Agent...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E8%AE%B0%E5%BF%86%E7%B3%BB%E7%BB%9F%E8%AE%BE%E8%AE%A1-%E7%9F%AD%E6%9C%9F-%E9%95%BF%E6%9C%9F%E4%B8%8E%E5%B7%A5%E4%BD%9C%E8%AE%B0%E5%BF%86/https://lingque.academy/learn/agent-engineering/agent%E8%AE%B0%E5%BF%86%E7%B3%BB%E7%BB%9F%E8%AE%BE%E8%AE%A1-%E7%9F%AD%E6%9C%9F-%E9%95%BF%E6%9C%9F%E4%B8%8E%E5%B7%A5%E4%BD%9C%E8%AE%B0%E5%BF%86/Agent 记忆系统设计：短期、长期与工作记忆 记忆架构（Buffer/Summary/Entity/Vector）、对话窗口管理、RAG 记忆检索与情景记忆实战 引言 人类的记忆系统分为短期记忆（工作记忆，容量约 7 项）、长期记忆（近乎无限容量）和情景记忆（特定事件的回忆）。LLM Agent 面临类似的记忆挑战：上下文窗口有限（类似工作记忆容量），需要在多轮对话和多次会话之间保持连续性。...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E8%AE%B0%E5%BF%86%E7%B3%BB%E7%BB%9F-%E4%BB%8E%E7%9F%AD%E6%9C%9F%E5%88%B0%E9%95%BF%E6%9C%9F/https://lingque.academy/learn/agent-engineering/agent%E8%AE%B0%E5%BF%86%E7%B3%BB%E7%BB%9F-%E4%BB%8E%E7%9F%AD%E6%9C%9F%E5%88%B0%E9%95%BF%E6%9C%9F/Agent 记忆系统：从短期到长期 构建具有持续记忆能力的 AI Agent 系统 为什么 Agent 需要记忆 大语言模型的原生状态是"无记忆"的——每次请求都是一次全新的开始。Agent 记忆系统的目标是弥补这一缺陷，让 Agent 具备： 会话连续性：在同一任务的多轮交互中保持上下文 跨会话持久性：记住上一次对话的结论和决策 经验积累：从历史任务中学习，避免重复犯错...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E8%AF%84%E4%BC%B0%E4%B8%8E%E5%9F%BA%E5%87%86%E6%B5%8B%E8%AF%95%E6%96%B9%E6%B3%95%E8%AE%BA/https://lingque.academy/learn/agent-engineering/agent%E8%AF%84%E4%BC%B0%E4%B8%8E%E5%9F%BA%E5%87%86%E6%B5%8B%E8%AF%95%E6%96%B9%E6%B3%95%E8%AE%BA/Agent 评估与基准测试方法论 如何系统化地衡量 AI Agent 的能力、可靠性与安全性 评估的核心挑战 Agent 评估比传统 ML 模型评估复杂得多，原因在于： 非确定性：相同输入可能产生不同的执行路径 多步骤：最终结果依赖中间每一步的质量 环境交互：Agent 的行为会改变环境状态，影响后续步骤 开放结局：很多任务没有唯一的"正确答案" 传统模型评估： 输入 ──→ 模型 ──→ 输出...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/agent%E8%AF%84%E6%B5%8B%E4%BD%93%E7%B3%BB-%E4%BB%8Eswe-bench%E5%88%B0%E8%87%AA%E5%AE%9A%E4%B9%89%E5%9F%BA%E5%87%86/https://lingque.academy/learn/agent-engineering/agent%E8%AF%84%E6%B5%8B%E4%BD%93%E7%B3%BB-%E4%BB%8Eswe-bench%E5%88%B0%E8%87%AA%E5%AE%9A%E4%B9%89%E5%9F%BA%E5%87%86/Agent 评测体系：从 SWE-bench 到自定义基准 为什么 Agent 评测不同于模型评测 模型评测关注的是"给定输入，输出是否正确"（如 MMLU、HumanEval）。Agent 评测则更复杂：Agent 的行为是多步骤的，涉及工具调用、环境交互、状态管理和错误恢复。同一个任务，Agent 可能通过完全不同的路径到达正确答案。 核心挑战： 非确定性：同一 Agent...Sat, 28 Feb 2026 00:00:00 GMTagent-engineeringhttps://lingque.academy/learn/agent-engineering/autogen%E6%B7%B1%E5%BA%A6%E8%A7%A3%E6%9E%90-%E5%BE%AE%E8%BD%AF%E5%A4%9Aagent%E6%A1%86%E6%9E%B6%E5%AE%9E%E6%88%98/https://lingque.academy/learn/agent-engineering/autogen%E6%B7%B1%E5%BA%A6%E8%A7%A3%E6%9E%90-%E5%BE%AE%E8%BD%AF%E5%A4%9Aagent%E6%A1%86%E6%9E%B6%E5%AE%9E%E6%88%98/AutoGen 深度解析：微软多 Agent 框架实战 概述 AutoGen 是微软研究院开源的多 Agent 对话框架，核心设计理念是通过 Agent 之间的对话来完成复杂任务。不同于单 Agent 的工具调用模式，AutoGen 让多个具有不同角色和能力的 Agent 协作对话，形成"群体智能"。 2024 年底 AutoGen 发布了 0.4 版本（AutoGen v0.4 /...Sat, 28 Feb 2026 00:00:00 GMTagent-engineering