OpenClaw 设计与实现

参考文献

作者 杨艺韬 · 1,722 字 · 发布于 · 更新于

本参考文献汇总了全专栏各章"📚 推荐阅读"中引用的所有资源，按章节组织；少数条目（第 16、18 章）为本页补充的相关资料。链接已于 2026-09-25 核对并更新为现址。

第 1 章 — 为什么需要 OpenClaw

  1. The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey — Tula Masterman 等，arXiv:2404.11584（2024 年 4 月）. https://arxiv.org/abs/2404.11584
  2. Build a Large Language Model (From Scratch) — Sebastian Raschka. Manning. https://www.manning.com/books/build-a-large-language-model-from-scratch
  3. OpenClaw GitHub 仓库. https://github.com/openclaw/openclaw
  4. ReAct: Synergizing Reasoning and Acting in Language Models — Shunyu Yao 等，arXiv:2210.03629（2022 年 10 月）. https://arxiv.org/abs/2210.03629
  5. OpenClaw 官方文档. https://docs.openclaw.ai/
  6. Awesome LLM Agents — GitHub 资源合集. https://github.com/kaushikb11/awesome-llm-agents

第 2 章 — 架构总览

  1. Designing Data-Intensive Applications — Martin Kleppmann. O'Reilly Media. https://dataintensive.net/
  2. Pattern-Oriented Software Architecture (POSA) — Buschmann et al. https://en.wikipedia.org/wiki/Pattern-Oriented_Software_Architecture
  3. The Architecture of Open Source Applications. https://aosabook.org/
  4. C4 Model for Software Architecture — Simon Brown. https://c4model.com/
  5. OpenClaw 官方文档（同第 5 条）. https://docs.openclaw.ai/

第 3 章 — Gateway 网关引擎

  1. Kong Gateway 文档. https://developer.konghq.com/gateway/
  2. Node.js 事件循环深入理解（The Node.js Event Loop）. https://nodejs.org/en/learn/asynchronous-work/event-loop-timers-and-nexttick
  3. The Twelve-Factor App — Adam Wiggins. https://12factor.net/

第 4 章 — Provider 抽象层

  1. LiteLLM — 统一 LLM API. https://github.com/BerriAI/litellm
  2. Claude API 文档（原 Anthropic API 文档，docs.anthropic.com 已迁至此）. https://platform.claude.com/docs/
  3. OpenAI Cookbook. https://developers.openai.com/cookbook
  4. OpenRouter. https://openrouter.ai/
  5. AI SDK (Vercel). https://ai-sdk.dev/

第 5 章 — Session 与对话管理

  1. MemGPT: Towards LLMs as Operating Systems — Charles Packer 等，arXiv:2310.08560（2023 年 10 月）. https://arxiv.org/abs/2310.08560
  2. Conversation Design — Google. https://developers.google.com/assistant/conversation-design/welcome
  3. LangChain Blog（博客首页；上下文窗口管理相关文章可在此检索，并非一篇同名文章）. https://www.langchain.com/blog

第 6 章 — Agent 系统

  1. A Survey on Large Language Model based Autonomous Agents — Lei Wang 等，arXiv:2308.11432（2023 年 8 月）. https://arxiv.org/abs/2308.11432
  2. AutoGPT. https://github.com/Significant-Gravitas/AutoGPT
  3. CrewAI 文档. https://docs.crewai.com/
  4. Tree of Thoughts: Deliberate Problem Solving with Large Language Models — Shunyu Yao 等，arXiv:2305.10601（2023 年 5 月）. https://arxiv.org/abs/2305.10601
  5. Microsoft AutoGen. https://github.com/microsoft/autogen

第 7 章 — 通道架构

  1. Enterprise Integration Patterns — Gregor Hohpe & Bobby Woolf. https://www.enterpriseintegrationpatterns.com/
  2. Telegram Bot API 文档. https://core.telegram.org/bots/api
  3. Adapter Pattern — Refactoring Guru. https://refactoring.guru/design-patterns/adapter

第 8 章 — 通道实现深度剖析

  1. Discord 开发者文档. https://docs.discord.com/developers/intro
  2. WhatsApp Business Platform — Meta. https://developers.facebook.com/documentation/business-messaging/whatsapp/overview
  3. Signal Protocol 技术文档. https://signal.org/docs/

第 9 章 — 插件与扩展系统

  1. VS Code Extension API. https://code.visualstudio.com/api
  2. Microkernel — Wikipedia. https://en.wikipedia.org/wiki/Microkernel
  3. Plugins in Semantic Kernel — Microsoft. https://learn.microsoft.com/en-us/semantic-kernel/concepts/plugins/
  4. OSGi 规范. https://www.osgi.org/
  5. Obsidian 插件开发文档. https://docs.obsidian.md/Plugins

第 10 章 — 工具系统

  1. Toolformer: Language Models Can Teach Themselves to Use Tools — Timo Schick 等，arXiv:2302.04761（2023 年 2 月）. https://arxiv.org/abs/2302.04761
  2. Claude Tool Use 文档 — Anthropic. https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview
  3. Playwright 自动化框架. https://playwright.dev/

第 11 章 — Node 系统与设备连接

  1. RFC 6455: The WebSocket Protocol — IETF. https://datatracker.ietf.org/doc/html/rfc6455
  2. Matter — Connectivity Standards Alliance (CSA). https://csa-iot.org/all-solutions/matter/
  3. How Tailscale Works — Tailscale. https://tailscale.com/blog/how-tailscale-works
  4. Web of Things — W3C. https://www.w3.org/WoT/
  5. Home Assistant 开发文档. https://developers.home-assistant.io/

第 12 章 — 定时任务与自动化

  1. OpenClaw CHANGELOG（v2026.3.31）. https://github.com/openclaw/openclaw/blob/v2026.3.31/CHANGELOG.md
  2. Croner — OpenClaw 实际使用的定时任务库（croner ^10.0.1）. https://github.com/hexagon/croner
  3. Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance — Yaxi Lu 等，arXiv:2410.12361（2024 年 10 月）. https://arxiv.org/abs/2410.12361
  4. Temporal.io — 工作流编排引擎. https://temporal.io/

第 13 章 — 安全与权限

  1. OWASP Top 10 for LLM Applications. https://owasp.org/www-project-top-10-for-large-language-model-applications/
  2. Prompt injection: What's the worst that can happen? — Simon Willison（2023 年 4 月）. https://simonwillison.net/2023/Apr/14/worst-that-can-happen/
  3. Least Privilege（NIST CSRC 术语表） — NIST. https://csrc.nist.gov/glossary/term/least_privilege
  4. Anthropic's Responsible Scaling Policy（Anthropic 负责任扩展政策）. https://www.anthropic.com/news/anthropics-responsible-scaling-policy
  5. LLM Guard. https://github.com/protectai/llm-guard

第 14 章 — CLI 与交互界面

  1. Command Line Interface Guidelines. https://clig.dev/
  2. Ink — React for CLI. https://github.com/vadimdemedes/ink
  3. The Art of Unix Programming — Eric S. Raymond. http://www.catb.org/~esr/writings/taoup/

第 15 章 — 部署与运维

  1. Docker 多阶段构建（Multi-stage builds）. https://docs.docker.com/build/building/multi-stage/
  2. The Twelve-Factor App — Adam Wiggins（同第 14 条）. https://12factor.net/
  3. Prometheus 监控实践. https://prometheus.io/docs/practices/

第 16 章 — 技能系统

  1. Retrieval-Augmented Generation for Large Language Models: A Survey — Yunfan Gao 等，arXiv:2312.10997（2023 年 12 月）. https://arxiv.org/abs/2312.10997
  2. Custom instructions for ChatGPT — OpenAI（2023 年 7 月）. https://openai.com/index/custom-instructions-for-chatgpt/
  3. Creating Skills — OpenClaw 官方文档中的技能编写指南（仓库内 docs/tools/creating-skills.md；本页补充）. https://docs.openclaw.ai/tools/creating-skills
  4. Model Context Protocol (MCP) — Anthropic 提出的开放协议. https://modelcontextprotocol.io/
  5. Semantic Kernel — Microsoft（早期称 Skills，现称 Plugins）. https://learn.microsoft.com/en-us/semantic-kernel/

第 17 章 — 设计模式与架构决策

  1. Design Patterns: Elements of Reusable Object-Oriented Software — Gamma, Helm, Johnson & Vlissides (GoF). https://en.wikipedia.org/wiki/Design_Patterns
  2. Software Architecture: The Hard Parts — Ford, Richards, Sadalage & Dehghani. O'Reilly Media, 2021. https://www.oreilly.com/library/view/software-architecture-the/9781492086888/
  3. Architectural Styles and the Design of Network-based Software Architectures — Roy Fielding (博士论文). https://ics.uci.edu/~fielding/pubs/dissertation/top.htm

第 18 章 — 构建你自己的 Agent 帝国

  1. Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations — Shoham & Leyton-Brown. https://www.masfoundations.org/
  2. The Society of Mind — Marvin Minsky. https://en.wikipedia.org/wiki/Society_of_Mind
  3. Claude Tool Use 文档 — Anthropic（同第 40 条）. https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview
  4. Building Effective Agents（Agent 编排模式；本页补充） — Anthropic. https://www.anthropic.com/engineering/building-effective-agents
  5. Agent Protocol — 最初由 AI Engineer Foundation 发起（仓库现由 AGI, Inc. 维护，https://github.com/agi-inc/agent-protocol）. https://agentprotocol.ai/
  6. LangGraph (LangChain). https://langchain-ai.github.io/langgraph/

资源分类索引

编号与上文正文一一对应（共 75 条，含 3 条同一资源在不同章的重复引用）。

学术论文（均已通过 arXiv 页面核对标题、第一作者与发表时间）

# 论文 第一作者 年份 相关章节
1 The Landscape of Emerging AI Agent Architectures (arXiv:2404.11584) Tula Masterman 2024 第 1 章
4 ReAct: Synergizing Reasoning and Acting in Language Models (arXiv:2210.03629) Shunyu Yao 2022 第 1 章
20 MemGPT: Towards LLMs as Operating Systems (arXiv:2310.08560) Charles Packer 2023 第 5 章
23 A Survey on Large Language Model based Autonomous Agents (arXiv:2308.11432) Lei Wang 2023 第 6 章
26 Tree of Thoughts (arXiv:2305.10601) Shunyu Yao 2023 第 6 章
39 Toolformer (arXiv:2302.04761) Timo Schick 2023 第 10 章
49 Proactive Agent: Shifting LLM Agents from Reactive Responses to Active Assistance (arXiv:2410.12361) Yaxi Lu 2024 第 12 章
62 Retrieval-Augmented Generation for LLMs: A Survey (arXiv:2312.10997) Yunfan Gao 2023 第 16 章
69 Architectural Styles and the Design of Network-based Software Architectures（博士论文） Roy Fielding 2000 第 17 章

经典著作

# 书名 作者 相关章节
2 Build a Large Language Model (From Scratch) Sebastian Raschka 第 1 章
7 Designing Data-Intensive Applications Martin Kleppmann 第 2 章
8 Pattern-Oriented Software Architecture Buschmann et al. 第 2 章
28 Enterprise Integration Patterns Hohpe & Woolf 第 7 章
58 The Art of Unix Programming Eric S. Raymond 第 14 章
67 Design Patterns (GoF) Gamma, Helm, Johnson & Vlissides 第 17 章
68 Software Architecture: The Hard Parts Ford, Richards, Sadalage & Dehghani 第 17 章
70 Multiagent Systems Shoham & Leyton-Brown 第 18 章
71 The Society of Mind Marvin Minsky 第 18 章

开源项目

# 项目 相关章节
3 OpenClaw 第 1 章
6 Awesome LLM Agents 第 1 章
15 LiteLLM 第 4 章
24 AutoGPT 第 6 章
27 Microsoft AutoGen 第 6 章
41 Playwright 第 10 章
48 Croner（OpenClaw 实际使用的定时任务库） 第 12 章
55 LLM Guard 第 13 章
57 Ink (React for CLI) 第 14 章
75 LangGraph 第 18 章

官方文档与标准

# 资源 相关章节
5 / 11 OpenClaw 官方文档 第 1、2 章
12 Kong Gateway 第 3 章
16 Claude API 第 4 章
29 Telegram Bot API 第 7 章
31 Discord 开发者文档 第 8 章
42 RFC 6455（WebSocket） 第 11 章
51 OWASP Top 10 for LLM Applications 第 13 章
56 Command Line Interface Guidelines 第 14 章
64 OpenClaw Creating Skills 第 16 章
65 Model Context Protocol 第 16 章

共计 75 条参考文献，其中 9 篇学术论文、9 部经典著作、10 个开源项目，其余为官方文档与技术资源。