[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-agent0ai--agent-zero":3,"tool-agent0ai--agent-zero":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":80,"owner_twitter":81,"owner_website":82,"owner_url":83,"languages":84,"stars":109,"forks":110,"last_commit_at":111,"license":112,"difficulty_score":10,"env_os":113,"env_gpu":114,"env_ram":114,"env_deps":115,"category_tags":118,"github_topics":119,"view_count":126,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":127,"updated_at":128,"faqs":129,"releases":158},1154,"agent0ai\u002Fagent-zero","agent-zero","Agent Zero AI framework","Agent Zero 是一个动态的、可成长的智能代理框架，旨在作为个人助手帮助完成各种任务。它不预设特定功能，而是根据用户的使用不断学习和进化，能够收集信息、执行命令、编写代码，并与其他代理协作解决问题。它将计算机视为工具，通过操作系统完成任务，具备持久记忆能力，能记住过往的经验以提高效率。\n\n适合开发者、研究人员以及希望提升工作效率的用户。其独特的技能系统采用开放标准，支持多种编程环境，便于扩展和定制。同时，Agent Zero 提供了对 Git 项目的支持，允许在安全的工作空间中操作代码库。","\u003Cdiv align=\"center\">\n\n# `Agent Zero`\n\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F11745\" target=\"_blank\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_4a68feb902da.png\" alt=\"frdel%2Fagent-zero | Trendshift\" style=\"width: 250px; height: 55px;\" width=\"250\" height=\"55\"\u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\n[![Agent Zero Website](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWebsite-agent--zero.ai-0A192F?style=for-the-badge&logo=vercel&logoColor=white)](https:\u002F\u002Fagent-zero.ai) [![Thanks to Sponsors](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGitHub%20Sponsors-Thanks%20to%20Sponsors-FF69B4?style=for-the-badge&logo=githubsponsors&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Fagent0ai) [![Follow on X](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-Follow-000000?style=for-the-badge&logo=x&logoColor=white)](https:\u002F\u002Fx.com\u002FAgent0ai) [![Join our Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20our%20server-5865F2?style=for-the-badge&logo=discord&logoColor=white)](https:\u002F\u002Fdiscord.gg\u002FB8KZKNsPpj) [![Subscribe on YouTube](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-Subscribe-red?style=for-the-badge&logo=youtube&logoColor=white)](https:\u002F\u002Fwww.youtube.com\u002F@AgentZeroFW) [![Connect on LinkedIn](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLinkedIn-Connect-blue?style=for-the-badge&logo=linkedin&logoColor=white)](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fjan-tomasek\u002F) [![Follow on Warpcast](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWarpcast-Follow-5A32F3?style=for-the-badge)](https:\u002F\u002Fwarpcast.com\u002Fagent-zero)\n\n\n## Documentation:\n\n[Introduction](#a-personal-organic-agentic-framework-that-grows-and-learns-with-you) •\n[Installation](.\u002Fdocs\u002Fsetup\u002Finstallation.md) •\n[How to update](.\u002Fdocs\u002Fsetup\u002Finstallation.md#how-to-update-agent-zero) \u003Cbr>\n[Development Setup](.\u002Fdocs\u002Fsetup\u002Fdev-setup.md) •\n[Usage](.\u002Fdocs\u002Fguides\u002Fusage.md)\n\nOr see DeepWiki generated documentation:\n\n[![Ask DeepWiki](https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg)](https:\u002F\u002Fdeepwiki.com\u002Fagent0ai\u002Fagent-zero)\n\n\u003C\u002Fdiv>\n\n\n\u003Cdiv align=\"center\">\n\n> ### 🚨 **AGENT ZERO SKILLS** 🚨\n> **Skills System** - portable, structured agent capabilities using the open `SKILL.md` standard (compatible with Claude Code, Codex and more).\n> \n> **Plus:** Git-based Projects with authentication for public\u002Fprivate repositories - clone codebases directly into isolated workspaces.\n> \n> See [Usage Guide](.\u002Fdocs\u002Fguides\u002Fusage.md) and [Projects Tutorial](.\u002Fdocs\u002Fguides\u002Fprojects.md) to get started.\n\u003C\u002Fdiv>\n\n\n\n[![Showcase](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_0fea71ff4bf3.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=MdzLhWWoxEs)\n\n\n## A personal, organic agentic framework that grows and learns with you\n\n\n\n- Agent Zero is not a predefined agentic framework. It is designed to be dynamic, organically growing, and learning as you use it.\n- Agent Zero is fully transparent, readable, comprehensible, customizable, and interactive.\n- Agent Zero uses the computer as a tool to accomplish its (your) tasks.\n\n# ⚙️ Installation\n\nClick to open a video to learn how to install Agent Zero:\n\n[![Easy Installation guide](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_05a7e84ddca3.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=2-qFNUvqrXA)\n\n### ⚡ Quick Start\n\n**macOS \u002F Linux:**\n```bash\ncurl -fsSL https:\u002F\u002Fbash.agent-zero.ai | bash\n```\n\n**Windows (PowerShell):**\n```powershell\nirm https:\u002F\u002Fps.agent-zero.ai | iex\n```\n\n**Docker:**\n```bash\ndocker run -p 80:80 agent0ai\u002Fagent-zero\n```\n\nA detailed setup guide for Windows, macOS, and Linux can be found in the Agent Zero Documentation at [this page](.\u002Fdocs\u002Fsetup\u002Finstallation.md).\n\n\n# 💡 Key Features\n\n1. **General-purpose Assistant**\n\n- Agent Zero is not pre-programmed for specific tasks (but can be). It is meant to be a general-purpose personal assistant. Give it a task, and it will gather information, execute commands and code, cooperate with other agent instances, and do its best to accomplish it.\n- It has a persistent memory, allowing it to memorize previous solutions, code, facts, instructions, etc., to solve tasks faster and more reliably in the future.\n\n![Agent 0 Working](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_1ea1b21f0376.png)\n\n2. **Computer as a Tool**\n\n- Agent Zero uses the operating system as a tool to accomplish its tasks. It has no single-purpose tools pre-programmed. Instead, it can write its own code and use the terminal to create and use its own tools as needed.\n- The only default tools in its arsenal are online search, memory features, communication (with the user and other agents), and code\u002Fterminal execution. Everything else is created by the agent itself or can be extended by the user.\n- Tool usage functionality has been developed from scratch to be the most compatible and reliable, even with very small models.\n- **Default Tools:** Agent Zero includes tools like knowledge, code execution, and communication.\n- **Creating Custom Tools:** Extend Agent Zero's functionality by creating your own custom tools.\n- **Skills (SKILL.md Standard):** Skills are contextual expertise loaded dynamically when relevant. They use the open SKILL.md standard (developed by Anthropic), making them compatible with Claude Code, Cursor, Goose, OpenAI Codex CLI, and GitHub Copilot.\n\n3. **Multi-agent Cooperation**\n\n- Every agent has a superior agent giving it tasks and instructions. Every agent then reports back to its superior.\n- In the case of the first agent in the chain (Agent 0), the superior is the human user; the agent sees no difference.\n- Every agent can create its subordinate agent to help break down and solve subtasks. This helps all agents keep their context clean and focused.\n\n![Multi-agent](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_03b40a2d4933.png)\n\n### Browser Agent\n\n- Browser automation is provided by the built-in `_browser_agent` plugin.\n- It uses the effective Main Model resolved by `_model_config`; there is no separate browser model slot.\n- Browser vision follows the Main Model's vision setting.\n- Playwright Chromium: **Docker** images ship the headless shell preinstalled. **Local development** installs it on first Browser Agent use via `ensure_playwright_binary()` in `plugins\u002F_browser_agent\u002Fhelpers\u002Fplaywright.py` (into `tmp\u002Fplaywright`); you can pre-install manually (see [Development Setup](docs\u002Fsetup\u002Fdev-setup.md)) to skip the wait.\n\n4. **Completely Customizable and Extensible**\n\n- Almost nothing in this framework is hard-coded. Nothing is hidden. Everything can be extended or changed by the user.\n- The whole behavior is defined by a system prompt in the **prompts\u002Fdefault\u002Fagent.system.md** file. Change this prompt and change the framework dramatically.\n- The framework does not guide or limit the agent in any way. There are no hard-coded rails that agents have to follow.\n- Every prompt, every small message template sent to the agent in its communication loop can be found in the **prompts\u002F** folder and changed.\n- Built-in tools live in the core **tools\u002F** folder or in built-in plugins under **plugins\u002F** and can be adapted or extended.\n- **Automated configuration** via `A0_SET_` environment variables for deployment automation and easy setup.\n\n![Prompts](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_977b4e7f16f2.png)\n\n5. **Communication is Key**\n\n- Give your agent a proper system prompt and instructions, and it can do miracles.\n- Agents can communicate with their superiors and subordinates, asking questions, giving instructions, and providing guidance. Instruct your agents in the system prompt on how to communicate effectively.\n- The terminal interface is real-time streamed and interactive. You can stop and intervene at any point. If you see your agent heading in the wrong direction, just stop and tell it right away.\n- There is a lot of freedom in this framework. You can instruct your agents to regularly report back to superiors asking for permission to continue. You can instruct them to use point-scoring systems when deciding when to delegate subtasks. Superiors can double-check subordinates' results and dispute. The possibilities are endless.\n\n## 🚀 Real-world use cases\n\n- **Financial Analysis & Charting** - `\"Find last month's Bitcoin\u002FUSD price trend, correlate with major cryptocurrency news events, generate annotated chart with highlighted key dates\"`\n\n- **Excel Automation Pipeline** - `\"Scan incoming directory for financial spreadsheets, validate and clean data, consolidate from multiple sources, generate executive reports with flagged anomalies\"`\n\n- **API Integration Without Code** - `\"Use this Google Gemini API snippet to generate product images, remember the integration for future use\"` - agent learns and stores the solution in memory\n\n- **Automated Server Monitoring** - `\"Check server status every 30 minutes: CPU usage, disk space, memory. Alert if metrics exceed thresholds\"` (scheduled task with project-scoped credentials)\n\n- **Multi-Client Project Isolation** - Separate projects for each client with isolated memory, custom instructions, and dedicated secrets - prevents context bleed across sensitive work\n\n## 🐳 Fully Dockerized, with Speech-to-Text and TTS\n\n![Settings](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_b2e4881fe05a.png)\n\n- Customizable settings allow users to tailor the agent's behavior and responses to their needs.\n- The Web UI output is very clean, fluid, colorful, readable, and interactive; nothing is hidden.\n- You can load or save chats directly within the Web UI.\n- The same output you see in the terminal is automatically saved to an HTML file in **logs\u002F** folder for every session.\n\n![Time example](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_a494b702185a.jpg)\n\n- Agent output is streamed in real-time, allowing users to read along and intervene at any time.\n- No coding is required; only prompting and communication skills are necessary.\n- With a solid system prompt, the framework is reliable even with small models, including precise tool usage.\n\n## 👀 Keep in Mind\n\n1. **Agent Zero Can Be Dangerous!**\n\n- With proper instruction, Agent Zero is capable of many things, even potentially dangerous actions concerning your computer, data, or accounts. Always run Agent Zero in an isolated environment (like Docker) and be careful what you wish for.\n\n2. **Agent Zero Is Prompt-based.**\n\n- The whole framework is guided by the **prompts\u002F** folder. Agent guidelines, tool instructions, messages, utility AI functions, it's all there.\n\n\n## 📚 Read the Documentation\n\n| Page | Description |\n|-------|-------------|\n| [Installation](.\u002Fdocs\u002Fsetup\u002Finstallation.md) | Installation, setup and configuration |\n| [Usage](.\u002Fdocs\u002Fguides\u002Fusage.md) | Basic and advanced usage |\n| [Guides](.\u002Fdocs\u002Fguides\u002F) | Step-by-step guides: Usage, Projects, API Integration, MCP Setup, A2A Setup |\n| [Development Setup](.\u002Fdocs\u002Fsetup\u002Fdev-setup.md) | Development and customization |\n| [WebSocket Infrastructure](.\u002Fdocs\u002Fdeveloper\u002Fwebsockets.md) | Real-time WebSocket handlers, client APIs, filtering semantics, envelopes |\n| [Extensions](.\u002Fdocs\u002Fdeveloper\u002Fextensions.md) | Extending Agent Zero |\n| [Connectivity](.\u002Fdocs\u002Fdeveloper\u002Fconnectivity.md) | External API endpoints, MCP server connections, A2A protocol |\n| [Architecture](.\u002Fdocs\u002Fdeveloper\u002Farchitecture.md) | System design and components |\n| [Contributing](.\u002Fdocs\u002Fguides\u002Fcontribution.md) | How to contribute |\n| [Troubleshooting](.\u002Fdocs\u002Fguides\u002Ftroubleshooting.md) | Common issues and their solutions |\n\n\n## 🤝 Community and Support\n\n- [Join our Discord](https:\u002F\u002Fdiscord.gg\u002FB8KZKNsPpj) for live discussions or [visit our Skool Community](https:\u002F\u002Fwww.skool.com\u002Fagent-zero).\n- [Follow our YouTube channel](https:\u002F\u002Fwww.youtube.com\u002F@AgentZeroFW) for hands-on explanations and tutorials\n- [Report Issues](https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues) for bug fixes and features\n","\u003Cdiv align=\"center\">\n\n# `Agent Zero`\n\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F11745\" target=\"_blank\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_4a68feb902da.png\" alt=\"frdel%2Fagent-zero | Trendshift\" style=\"width: 250px; height: 55px;\" width=\"250\" height=\"55\"\u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\n[![Agent Zero Website](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWebsite-agent--zero.ai-0A192F?style=for-the-badge&logo=vercel&logoColor=white)](https:\u002F\u002Fagent-zero.ai) [![Thanks to Sponsors](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGitHub%20Sponsors-Thanks%20to%20Sponsors-FF69B4?style=for-the-badge&logo=githubsponsors&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Fagent0ai) [![Follow on X](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-Follow-000000?style=for-the-badge&logo=x&logoColor=white)](https:\u002F\u002Fx.com\u002FAgent0ai) [![Join our Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20our%20server-5865F2?style=for-the-badge&logo=discord&logoColor=white)](https:\u002F\u002Fdiscord.gg\u002FB8KZKNsPpj) [![Subscribe on YouTube](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FYouTube-Subscribe-red?style=for-the-badge&logo=youtube&logoColor=white)](https:\u002F\u002Fwww.youtube.com\u002F@AgentZeroFW) [![Connect on LinkedIn](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLinkedIn-Connect-blue?style=for-the-badge&logo=linkedin&logoColor=white)](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fjan-tomasek\u002F) [![Follow on Warpcast](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWarpcast-Follow-5A32F3?style=for-the-badge)](https:\u002F\u002Fwarpcast.com\u002Fagent-zero)\n\n\n## 文档：\n\n[简介](#a-personal-organic-agentic-framework-that-grows-and-learns-with-you) •\n[安装](.\u002Fdocs\u002Fsetup\u002Finstallation.md) •\n[如何更新](.\u002Fdocs\u002Fsetup\u002Finstallation.md#how-to-update-agent-zero) \u003Cbr>\n[开发环境搭建](.\u002Fdocs\u002Fsetup\u002Fdev-setup.md) •\n[使用指南](.\u002Fdocs\u002Fguides\u002Fusage.md)\n\n或者查看由 DeepWiki 生成的文档：\n\n[![Ask DeepWiki](https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg)](https:\u002F\u002Fdeepwiki.com\u002Fagent0ai\u002Fagent-zero)\n\n\u003C\u002Fdiv>\n\n\n\u003Cdiv align=\"center\">\n\n> ### 🚨 **AGENT ZERO 技能** 🚨\n> **技能系统** - 基于开放 `SKILL.md` 标准的可移植、结构化的智能体能力（与 Claude Code、Codex 等兼容）。\n> \n> **此外：** 基于 Git 的项目，支持公共和私有仓库的身份验证——可将代码库直接克隆到隔离的工作空间中。\n> \n> 请参阅 [使用指南](.\u002Fdocs\u002Fguides\u002Fusage.md) 和 [项目教程](.\u002Fdocs\u002Fguides\u002Fprojects.md) 开始使用。\n\u003C\u002Fdiv>\n\n\n\n[![展示](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_0fea71ff4bf3.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=MdzLhWWoxEs)\n\n\n## 一个与你共同成长、学习的个性化有机智能体框架\n\n\n\n- Agent Zero 并非预定义的智能体框架。它被设计成动态的、有机生长的，并且会随着你的使用不断学习。\n- Agent Zero 完全透明、可读、易懂、可定制且高度互动。\n- Agent Zero 将计算机作为工具来完成其（你的）任务。\n\n# ⚙️ 安装\n\n点击观看视频，了解如何安装 Agent Zero：\n\n[![简易安装指南](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_05a7e84ddca3.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=2-qFNUvqrXA)\n\n### ⚡ 快速入门\n\n**macOS \u002F Linux:**\n```bash\ncurl -fsSL https:\u002F\u002Fbash.agent-zero.ai | bash\n```\n\n**Windows (PowerShell):**\n```powershell\nirm https:\u002F\u002Fps.agent-zero.ai | iex\n```\n\n**Docker:**\n```bash\ndocker run -p 80:80 agent0ai\u002Fagent-zero\n```\n\n关于 Windows、macOS 和 Linux 的详细安装指南，请参阅 Agent Zero 文档中的 [此页面](.\u002Fdocs\u002Fsetup\u002Finstallation.md)。\n\n\n# 💡 核心功能\n\n1. **通用型助手**\n\n- Agent Zero 并非为特定任务预先编程（但可以）。它旨在成为一个通用的个人助手。只需赋予它一项任务，它便会收集信息、执行命令和代码、与其他智能体实例协作，并尽最大努力完成任务。\n- 它拥有持久化记忆，能够记住之前的解决方案、代码、事实、指令等，从而在未来更快、更可靠地解决任务。\n\n![Agent 0 工作场景](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_1ea1b21f0376.png)\n\n2. **计算机作为工具**\n\n- Agent Zero 将操作系统用作完成任务的工具。它没有预置任何单一用途的工具。相反，它可以自行编写代码，并利用终端创建和使用所需的工具。\n- 其工具库中仅包含在线搜索、记忆功能、与用户及其他智能体沟通以及代码\u002F终端执行等默认工具。其他一切功能均由智能体自身创建，或可由用户扩展。\n- 工具使用功能从零开始开发，旨在实现最高兼容性和可靠性，即使在非常小的模型上也能稳定运行。\n- **默认工具：** Agent Zero 包含知识检索、代码执行和通信等功能。\n- **自定义工具：** 通过创建自己的自定义工具，进一步扩展 Agent Zero 的功能。\n- **技能（SKILL.md 标准）：** 技能是根据上下文动态加载的专业知识。它们采用 Anthropic 开发的开放 SKILL.md 标准，因此与 Claude Code、Cursor、Goose、OpenAI Codex CLI 和 GitHub Copilot 等工具兼容。\n\n3. **多智能体协作**\n\n- 每个智能体都有一个上级智能体为其分配任务和指令。每个智能体随后会向其上级汇报工作进展。\n- 对于链条中的第一个智能体（Agent 0）而言，其上级就是人类用户；该智能体对此并无感知差异。\n- 每个智能体都可以创建下属智能体，以帮助分解并解决子任务。这有助于所有智能体保持清晰、专注的工作上下文。\n\n![多智能体协作](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_03b40a2d4933.png)\n\n### 浏览器代理\n\n- 浏览器自动化由内置的 `_browser_agent` 插件提供。\n- 它使用由 `_model_config` 解析的有效主模型；没有单独的浏览器模型槽位。\n- 浏览器视觉遵循主模型的视觉设置。\n- Playwright Chromium：**Docker** 镜像预装了无头 Shell。**本地开发**会在首次使用浏览器代理时，通过 `plugins\u002F_browser_agent\u002Fhelpers\u002Fplaywright.py` 中的 `ensure_playwright_binary()` 方法将其安装到 `tmp\u002Fplaywright` 目录中；你也可以手动预先安装（参见 [开发设置](docs\u002Fsetup\u002Fdev-setup.md)），以避免等待。\n\n4. **完全可定制和可扩展**\n\n- 该框架中几乎没有硬编码的部分。没有任何隐藏的内容。用户可以扩展或更改任何部分。\n- 整个行为由 **prompts\u002Fdefault\u002Fagent.system.md** 文件中的系统提示定义。修改此提示即可显著改变框架的行为。\n- 框架不会以任何方式引导或限制代理。不存在代理必须遵循的硬编码规则。\n- 在代理的通信循环中发送的每一个提示、每一个小消息模板都可以在 **prompts\u002F** 文件夹中找到并进行修改。\n- 内置工具位于核心的 **tools\u002F** 文件夹中，或位于 **plugins\u002F** 下的内置插件中，可以被调整或扩展。\n- 通过 `A0_SET_` 环境变量实现**自动化配置**，便于部署自动化和快速设置。\n\n![提示](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_977b4e7f16f2.png)\n\n5. **沟通是关键**\n\n- 为你的代理提供合适的系统提示和指令，它就能创造奇迹。\n- 代理可以与其上级和下级进行沟通，提出问题、下达指令并提供指导。在系统提示中指导你的代理如何有效地进行沟通。\n- 终端界面是实时流式传输且交互式的。你可以在任何时候停止并介入。如果你发现代理正朝着错误的方向发展，只需立即停止并告知它。\n- 该框架提供了极大的自由度。你可以指示代理定期向其上级汇报，请求继续执行的许可。你也可以指示它们在决定何时委派子任务时使用积分系统。上级可以对下级的结果进行复核并提出异议。可能性是无限的。\n\n## 🚀 实际应用场景\n\n- **财务分析与图表绘制** - `\"查找上个月比特币\u002F美元的价格走势，与主要加密货币新闻事件相关联，生成带有高亮关键日期的注释图表\"`\n\n- **Excel 自动化流水线** - `\"扫描传入目录中的财务电子表格，验证并清理数据，从多个来源整合数据，生成带有异常标记的高管报告\"`\n\n- **无需代码的 API 集成** - `\"使用这段 Google Gemini API 代码片段生成产品图片，记住这次集成以便日后使用\"` - 代理会学习并将解决方案存储在内存中。\n\n- **自动化服务器监控** - `\"每 30 分钟检查一次服务器状态：CPU 使用率、磁盘空间、内存。如果指标超过阈值，则发出警报\"`（具有项目范围凭据的计划任务）\n\n- **多客户项目隔离** - 为每个客户创建独立的项目，拥有隔离的内存、自定义指令和专用密钥——防止敏感工作之间的上下文泄露。\n\n## 🐳 完全 Docker 化，支持语音转文本和 TTS\n\n![设置](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_b2e4881fe05a.png)\n\n- 可定制的设置允许用户根据自身需求调整代理的行为和响应。\n- Web UI 的输出非常简洁、流畅、色彩丰富、易读且具有交互性；没有任何隐藏的内容。\n- 你可以直接在 Web UI 中加载或保存聊天记录。\n- 你在终端中看到的相同输出会自动保存到 **logs\u002F** 文件夹中的 HTML 文件中，每次会话都会保存一份。\n\n![时间示例](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_readme_a494b702185a.jpg)\n\n- 代理的输出是实时流式的，允许用户随时阅读并进行干预。\n- 无需编写代码，只需要具备提示和沟通技巧即可。\n- 只要有扎实的系统提示，即使使用小型模型，该框架也能可靠运行，包括精确地使用工具。\n\n## 👀 请注意\n\n1. **Agent Zero 可能存在危险！**\n\n- 在适当的指导下，Agent Zero 能够完成许多事情，甚至可能涉及对你的计算机、数据或账户的潜在危险操作。请务必在隔离环境中（如 Docker）运行 Agent Zero，并小心你的期望。\n\n2. **Agent Zero 基于提示运行。**\n\n- 整个框架都由 **prompts\u002F** 文件夹中的提示所引导。代理的指导方针、工具说明、消息以及实用的 AI 功能，全部都在这里。\n\n\n## 📚 阅读文档\n\n| 页面 | 描述 |\n|-------|-------------|\n| [安装](.\u002Fdocs\u002Fsetup\u002Finstallation.md) | 安装、设置和配置 |\n| [使用](.\u002Fdocs\u002Fguides\u002Fusage.md) | 基本及高级用法 |\n| [指南](.\u002Fdocs\u002Fguides\u002F) | 分步指南：使用、项目、API 集成、MCP 设置、A2A 设置 |\n| [开发设置](.\u002Fdocs\u002Fsetup\u002Fdev-setup.md) | 开发和定制 |\n| [WebSocket 基础设施](.\u002Fdocs\u002Fdeveloper\u002Fwebsockets.md) | 实时 WebSocket 处理程序、客户端 API、过滤语义、封装 |\n| [扩展](.\u002Fdocs\u002Fdeveloper\u002Fextensions.md) | 扩展 Agent Zero |\n| [连接性](.\u002Fdocs\u002Fdeveloper\u002Fconnectivity.md) | 外部 API 端点、MCP 服务器连接、A2A 协议 |\n| [架构](.\u002Fdocs\u002Fdeveloper\u002Farchitecture.md) | 系统设计和组件 |\n| [贡献](.\u002Fdocs\u002Fguides\u002Fcontribution.md) | 如何贡献 |\n| [故障排除](.\u002Fdocs\u002Fguides\u002Ftroubleshooting.md) | 常见问题及其解决方案 |\n\n\n## 🤝 社区和支持\n\n- 加入我们的 Discord 社区（https:\u002F\u002Fdiscord.gg\u002FB8KZKNsPpj）进行实时讨论，或访问我们的 Skool 社区（https:\u002F\u002Fwww.skool.com\u002Fagent-zero）。\n- 关注我们的 YouTube 频道（https:\u002F\u002Fwww.youtube.com\u002F@AgentZeroFW）以获取实践讲解和教程。\n- 如需修复 bug 或添加功能，请提交问题至 GitHub（https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues）。","# Agent Zero 快速上手指南\n\n## 环境准备\n\n| 项目 | 要求 |\n|---|---|\n| 操作系统 | macOS \u002F Linux \u002F Windows 10+ |\n| Docker | 20.10+（推荐，隔离更安全） |\n| Python | 3.10+（本地开发时需要） |\n| 网络 | 可访问 GitHub、Docker Hub |\n\n> 国内用户建议使用 [清华镜像](https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fhelp\u002Fdocker-ce\u002F) 加速 Docker 安装。\n\n## 安装步骤\n\n### 1. 一键脚本（推荐）\n\n**macOS \u002F Linux**\n```bash\ncurl -fsSL https:\u002F\u002Fbash.agent-zero.ai | bash\n```\n\n**Windows（PowerShell 管理员）**\n```powershell\nirm https:\u002F\u002Fps.agent-zero.ai | iex\n```\n\n### 2. Docker 方式（最干净）\n```bash\ndocker run -p 80:80 agent0ai\u002Fagent-zero\n```\n浏览器访问 http:\u002F\u002Flocalhost 即可。\n\n### 3. 国内镜像（可选）\n```bash\n# 使用阿里云镜像加速\ndocker run -p 80:80 registry.cn-hangzhou.aliyuncs.com\u002Fagent0ai\u002Fagent-zero\n```\n\n## 基本使用\n\n1. 启动后，在终端或 Web UI 输入第一条指令：\n   ```\n   帮我创建一个 Python 脚本，列出当前目录下所有 txt 文件\n   ```\n\n2. 观察 Agent 实时输出，可随时按 `Ctrl+C` 中断并补充说明。\n\n3. 任务完成后，Agent 会将脚本保存到本地，并告知路径。\n\n4. 进阶：创建项目隔离环境\n   ```\n   \u002Fproject create my_finance\n   ```\n   之后所有操作均在 `my_finance` 独立工作区，避免交叉污染。\n\n> 首次使用浏览器自动化时，Agent 会自动下载 Playwright 浏览器内核，国内用户可提前设置 `PLAYWRIGHT_DOWNLOAD_HOST=https:\u002F\u002Fcdn.npmmirror.com\u002Fbinaries\u002Fplaywright` 加速。","一家 5 人初创团队正在开发一款跨平台桌面应用，需要在两周内完成 MVP 并上线，团队里只有 1 名全栈工程师。\n\n### 没有 agent-zero 时\n- 工程师每天花 3-4 小时手动拉取 GitHub 私有仓库、切分支、合并 PR，CI\u002FCD 脚本反复调试，时间被碎片化。\n- 每次需求变更都要手动写 Electron + React 的打包脚本，跨平台测试靠本地虚拟机，出包慢且容易遗漏平台差异。\n- 产品经理想快速验证新功能，只能排队等工程师空出时间，沟通成本高，迭代节奏被拖慢。\n- 代码规范、依赖更新、漏洞扫描全靠人工检查，漏掉一次就导致线上崩溃，团队士气受挫。\n\n### 使用 agent-zero 后\n- agent-zero 一键克隆私有仓库到隔离工作区，自动识别分支策略并执行合并、打 tag，工程师把 3-4 小时压缩到 10 分钟。\n- 工程师用自然语言告诉 agent-zero“为 macOS、Windows、Linux 生成 Electron 打包脚本并跑通 GitHub Actions”，它即刻生成并推送 CI 配置，当晚就拿到三平台安装包。\n- 产品经理直接在 agent-zero 聊天框里描述“加一个暗黑模式开关”，它自动改代码、跑测试、发预览链接，当天即可灰度发布。\n- agent-zero 每晚定时扫描依赖、跑 lint、执行安全审计，第二天早晨把报告和修复 PR 推到 Slack，团队安心睡觉。\n\nagent-zero 让 1 名工程师像 5 人 DevOps 团队一样高效，两周内准时上线 MVP。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagent0ai_agent-zero_0fea71ff.png","agent0ai","Agent Zero","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fagent0ai_e0ffb39d.png","Agent Zero AI",null,"pr@agent-zero.ai","Agent0ai","https:\u002F\u002Fagent-zero.ai","https:\u002F\u002Fgithub.com\u002Fagent0ai",[85,89,93,97,101,105],{"name":86,"color":87,"percentage":88},"Python","#3572A5",52,{"name":90,"color":91,"percentage":92},"HTML","#e34c26",23.6,{"name":94,"color":95,"percentage":96},"JavaScript","#f1e05a",20.2,{"name":98,"color":99,"percentage":100},"CSS","#663399",3.8,{"name":102,"color":103,"percentage":104},"Shell","#89e051",0.3,{"name":106,"color":107,"percentage":108},"Dockerfile","#384d54",0.1,16744,3440,"2026-04-05T11:18:52","NOASSERTION","Linux, macOS, Windows","未说明",{"notes":116,"python":114,"dependencies":117},"支持 Docker 一键运行；本地开发首次使用浏览器代理时会自动安装 Playwright Chromium；可通过 A0_SET_ 环境变量实现自动化配置；建议始终在隔离环境（如 Docker）中运行以避免潜在风险",[],[13,15,14],[120,121,122,123,124,125],"agent","ai","assistant","autonomous","linux","zero",4,"2026-03-27T02:49:30.150509","2026-04-06T06:44:19.577755",[130,135,140,145,150,154],{"id":131,"question_zh":132,"answer_zh":133,"source_url":134},5215,"如何解决 Playwright 浏览器代理初始化失败的问题？","当使用 browser-agent 时，如果遇到 Playwright 初始化失败的错误，请运行命令 `playwright install` 来下载所需的浏览器。此外，确保已正确安装 Playwright，并且路径配置无误。","https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues\u002F312",{"id":136,"question_zh":137,"answer_zh":138,"source_url":139},5216,"如何解决 'Socket is closed' 错误？","在使用 code_execution_tool 进行长时间操作时，可能会出现 'Socket is closed' 错误。建议切换到本地 Python TTY 模式运行 Agent Zero，以避免此类问题。另外，确保使用的版本为 v0.9.4 或更高版本。","https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues\u002F648",{"id":141,"question_zh":142,"answer_zh":143,"source_url":144},5217,"如何解决 Agent 被困在重复发送相同消息的循环中？","如果 Agent 不断重复发送相同的消息，可以尝试更换更强大的模型（如 Gemini Flash 3）或调整上下文长度。此外，确保模型配置正确，避免因模型理解不足导致的循环行为。","https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues\u002F320",{"id":146,"question_zh":147,"answer_zh":148,"source_url":149},5218,"如何在 Windows 上正确运行 Ollama？","在 Windows 上运行 Ollama，可以通过 WSL 或直接在 Windows 环境下设置环境变量并启动服务。例如：`export OLLAMA_HOST=0.0.0.0` 后运行 `ollama serve`。","https:\u002F\u002Fgithub.com\u002Fagent0ai\u002Fagent-zero\u002Fissues\u002F590",{"id":151,"question_zh":152,"answer_zh":153,"source_url":134},5219,"如何解决浏览器代理无法提取 Hacker News 新闻的问题？","如果 browser-agent 在访问 Hacker News 时出错，首先确认是否已通过 `playwright install` 安装了所有依赖。若问题依旧，可尝试升级 Playwright 到最新版本，或检查代码中的路径配置是否正确。",{"id":155,"question_zh":156,"answer_zh":157,"source_url":134},5220,"如何修复 'NoneType' 错误？","当使用 browser-agent 时，如果遇到 NoneType 错误，可能是由于代码中某些函数未正确处理返回值。可以尝试更新 Agent Zero 到最新版本，或者手动修改 `\u002Fa0\u002Fpython\u002Fhelpers\u002Ftools.py` 文件中的相关代码逻辑。",[159,164,169,174,179,184,189,194,199,204,209,214,219,224,229,234,239,244,249,254],{"id":160,"version":161,"summary_zh":162,"released_at":163},104743,"v1.7","## Agent Zero – Release Notes\n\n### Prompt & Model Guidance Overhaul\n\n- Adopted compact default prompts across the core stack, plugin-owned prompts, and the agent0 overlay — default assembled prompt now sits under ~3 000 tokens while retaining all key guardrails\n- Restored essential tool-call JSON examples, communication\u002Fsolving\u002Ftips guidance, and the agent0 profile after earlier over-aggressive trimming caused regressions with frontier models\n- Added strict JSON guidance to steer models toward treating the closing brace as an end-of-sequence signal, and defaulted response style to concise with expansion on demand\n- Removed the obsolete `a0_small` profile; tool-call knowledge reference renamed to a generic framework file\n- Added a regression test that verifies the assembled default prompt stays within the token budget and retains critical guardrails\n\n### Early Tool Dispatch via Streaming JSON Detection\n\n- Tool calls are now dispatched as soon as the first fully closed top-level JSON object is detected in the model stream — the stream is stopped at that point rather than waiting for all remaining tokens\n- `DirtyJson` parser gains a `completed` flag that signals when the root structure is explicitly closed (not merely at end-of-file), enabling safe early extraction\n- Extraction is restricted to brace-delimited objects, matching the shape of all tool calls\n\n### Plugin Discovery Cards on Welcome Screen & Onboarding\n\n- A new always-enabled `_discovery` plugin adds a discovery surface to the welcome\u002Fdashboard screen, surfacing the Plugin Hub and integrations (Telegram, Email, WhatsApp) via feature cards\n- Cards include persistent dismiss\u002Frestore state and CTA routing to the relevant plugin config screens\n- Discovery cards are integrated into the final onboarding step so new users see integration opportunities immediately after setup\n- Cards are hidden while the missing-API-key onboarding banner is visible to reduce noise during initial configuration\n- Implemented entirely through the existing WebUI extension mechanism with no core welcome-screen changes\n\n### Bug Fixes\n\n- Fixed a crash during config retrieval when a plugin directory cannot be found — `get_plugin_config` and `get_default_plugin_config` now return early instead of passing a `None` path to the filesystem layer","2026-04-03T15:32:20",{"id":165,"version":166,"summary_zh":167,"released_at":168},104744,"v1.6","- **WhatsApp integration plugin** — New plugin connecting Agent Zero to WhatsApp via a Baileys JS bridge with QR\u002Fpairing-code setup. Supports self-chat and dedicated (bot account) modes, group chat with mention detection, quoted replies, media\u002Fattachment handling, typing indicators, Markdown-to-WhatsApp formatting, and account disconnect\u002Fswitch.\n- **WhatsApp: self-chat is now the default mode** — Default changed from dedicated to self-chat; a warning banner is shown when `allowed_numbers` is empty in this mode.\n- **WhatsApp: allowed-numbers filtering moved to Python** — Phone number filtering now happens in the Python poll handler (reading config fresh each cycle) rather than the JS bridge, fixing a bug where LID-based matching never worked.\n- **WhatsApp: improved bridge reliability** — Corrupt or missing `node_modules` are detected and automatically reinstalled; the poll loop stops after 5 consecutive bridge failures instead of spamming errors; Node.js absence halts the loop immediately; bridge restarts automatically when configuration changes.\n- **WhatsApp: storage paths centralized** — Session and media files moved from `usr\u002Fwhatsapp` to `tmp\u002Fwhatsapp` via a new `storage_paths.py` helper; paths are now consistent across all modules.\n- **Settings: API keys no longer overwritten on save** — Saving general settings no longer clobbers API keys that were set via the Configure Models dialog.\n- **Settings: model config UI fixes** — Restored provider fallback and empty option for utility preset; extracted a reusable model-field component and unified API key lifecycle handling.\n- **UI server restructuring** — Server setup extracted into a `UiServerRuntime` class with modular initialization; environment config, route registration, and transport handlers moved to `helpers\u002Fui_server.py`.\n- **Update system enhancements** — Added `released_at` timestamp tracking for git tags and branch heads; new `get_current_major_main_latest_info` finds the latest same-major version on the main branch; `major_upgrade_versions` and `main_branch_latest` fields added to the update info payload.","2026-03-31T20:16:48",{"id":170,"version":171,"summary_zh":172,"released_at":173},104745,"v1.5","## WebSocket System Rework & General Improvements\n\n- **WebSocket architecture rewrite** — New `WsHandler` base class, `WsManager` for connection tracking, event routing, and buffering; reduces WebSocket init from ~170 lines to ~10. Frontend adapter updated with exponential backoff, envelope validation, 50 MB cap, and improved handler registration.\n- **WebSocket security fix** — `dispatch_to_all_sids` snapshot now taken inside the lock to prevent security check bypass during concurrent disconnects.\n- **WebSocket memory leak fixes** — Resolved SID retention after disconnection and cleaned up unreferenced broadcast tasks.\n- **API\u002FWebSocket caching** — WebSocket and API caching parameters are now enabled.\n- **`a0_small` prompt set** — New compact prompt set for small models, reducing initial token usage from ~10k to ~2.5k tokens. Includes stricter tool-call guardrails, canonical JSON examples, exact tool-name matching, and memory usage hints.\n- **Fix: non-ASCII filename uploads** — Resolved a 500 error when uploading images with non-ASCII filenames via `\u002Fapi\u002Fimage_get`; adds graceful fallback for path handling in both development and Docker environments.\n- **Fix: agents self-correct tool call format** — Agents now attempt to fix their own malformed tool calls instead of immediately erroring.\n- **UI: chat navigation buttons** — Repositioned as a floating overlay in the bottom-right of the chat area.\n- **Log sanitization** — Printed output and HTML logs are now sanitized to handle invalid Unicode (including lone surrogates); log files opened with `utf-8` and `errors='replace'`.\n- **Plugin installer** — Centralized popularity checking with a configurable `POPULAR_PLUGIN_MIN_STARS` threshold.\n- **Onboarding plugin** — Added generated thumbnail.","2026-03-30T18:17:48",{"id":175,"version":176,"summary_zh":177,"released_at":178},104746,"v1.4","## Agent Zero – Release Notes\n\n**Onboarding, plugin hub improvements, and model config enhancements**\n\n- **Onboarding flow for missing API keys** – A new onboarding experience guides users when no API key is configured. A banner in the chat composer surfaces the missing-key status directly, replacing a dedicated backend status endpoint.\n- **Model config presets** – A model switcher component is now available in the chat input area, with support for active preset\u002Fmodel retrieval from the model config store.\n- **Agent progress in chat input** – Agent progress is shown as ghost text in the chat input placeholder for better at-a-glance awareness.\n- **Parallel plugin scans** – The serialized scan queue has been removed; plugin scans now run in parallel, each in its own temporary chat context. The \"queued\" state has been removed from the UI and store.\n- **Sanitized plugin markdown rendering** – Plugin READMEs and docs are now rendered through a shared safe-markdown pipeline using DOMPurify, preventing unsafe HTML from reaching the page. GitHub README link and image rebasing (including `releases` routes) is now centralized and applied consistently across the Plugin Hub, installed plugin README, and markdown modal docs.\n- **Fixed plugin README image paths** – README images and links are correctly rebased to GitHub `blob`\u002F`raw` paths, fixing broken images in plugin documentation.","2026-03-30T07:48:47",{"id":180,"version":181,"summary_zh":182,"released_at":183},104747,"v1.3","## Agent Zero – Release Notes\n\n- **Chat compaction plugin**: New built-in plugin that compacts chat history, helping manage long conversations and context window usage. Includes plugin thumbnail.\n- **WebUI `page-head` extension point**: Plugins can now inject `\u003Cscript>` and `\u003Clink>` tags directly into `\u003Chead>` via a new `x-extension` injection point in `index.html`, enabling cleaner loading of vendor libraries (e.g. mermaid.js, custom fonts) without DOM manipulation hacks.\n- **Plugin index fetch cleanup**: Removed redundant `force` parameter from the server-side plugin index fetch handler; cache-busting is now handled entirely client-side.\n- **Updated `a0-development` skill**.\n- **Release workflow improvements**: Added fallback environment variable support (`OPENROUTER_MODEL_NAME` \u002F `OPENROUTER_MODEL`), improved error handling with graceful fallback messaging, and relaxed branch\u002Ftag constraints in the Docker publish workflow.","2026-03-27T13:04:29",{"id":185,"version":186,"summary_zh":187,"released_at":188},104748,"v1.2","- **Plugin installer**: Added force-refresh support with cache-busting headers and request sequence tracking to prevent race conditions during concurrent index loads. New `openIndexView` and `reloadIndex` store methods enable explicit refresh control.\n- **Model config**: Moved models summary state from component into the store, with loading and caching support. Added `refreshModelsSummary` and `ensureModelsSummaryLoaded` helpers.\n- **Docker publish workflow**: Release notes are now generated dynamically via OpenRouter instead of being read from static markdown files. The `docs\u002Frelease_notes\u002F` directory and its versioned files have been removed; the workflow now requires `OPENROUTER_API_KEY` and `OPENROUTER_MODEL` environment variables and uses a system prompt template at `scripts\u002Fopenrouter_release_notes_system_prompt.md`.\n\n**Breaking changes**\n\n- The `docs\u002Frelease_notes\u002F` directory and its static release note files (`v1.0`, `v1.1`) have been removed. Any external tooling or workflows that referenced `RELEASE_NOTES_DIR` or those files must be updated to use the new dynamic generation approach.","2026-03-26T19:46:44",{"id":190,"version":191,"summary_zh":192,"released_at":193},104749,"v1.1","# v1.1\n\nCovers changes from the `v0.9.8` series through the current `v1.5` state, including the `v0.9.9` work that landed along the way.\n\n- Agent Zero now has a real plugin platform: developers can build and ship functionality independently of the core, with a large refactor that makes more of the framework extensible through plugins, hooks, and user-side development instead of core-only changes.\n- The new Plugin Hub brings an app-store-like workflow for discovery, installation, configuration, validation, scanning, README viewing, thumbnails, scoped toggles, and safer plugin management.\n- Core extensibility was expanded across prompts, tools, secrets, APIs, model providers, lifecycle hooks, file watching, and other internal systems, making custom integrations and deeper framework extensions much easier to build and maintain.\n- New communication integrations: built-in Email and Telegram plugins with config UIs, attachment handling, dispatcher and model routing, notifications, group-chat support, and a long round of reliability fixes.\n- Model and prompt architecture upgrades: model presets and search, per-chat model switching, browser-agent model simplification, OpenRouter and provider fixes, the PromptInclude plugin, and related configuration improvements.\n- Self-update and release-flow improvements: native Git-based self-update, stronger tag validation and remote tag fetching, better disconnect handling in the update UI, and the simplified release tag format from `vX.Y.Z` to `vX.Y`.\n- UI and chat improvements: chat-branching fixes, expanded `ALL`-mode responses, taller chat input, sidebar and header cleanup, quick-action redesign, plugin and settings modal polish, and more stable WebSocket and frontend cache behavior.\n- Security and reliability hardening: CSRF fixes for HTTPS and Chromium, dependency security pins, secret masking in code-execution output, safer plugin warnings, cross-device file move fixes, improved email whitelist handling, and cleaner retry and exception handling.\n- Core framework refactors: broad Python, API, and helper cleanup, extension-system rework, routing and caching improvements, generic tool-output update hooks, and extraction of more capabilities into built-in plugins.\n- Ongoing polish across the project: welcome-banner and default-preset cleanup, browse and extract improvements, built-in plugin branding and docs refreshes, project-doc fixes, and a fix for `skills_tool` loading when `loaded_skills` is uninitialized.","2026-03-26T15:32:26",{"id":195,"version":196,"summary_zh":197,"released_at":198},104750,"v0.9.8.3","### v0.9.8.3 - Improved chatbox expansion\r\n\r\n### v0.9.8.2 - Bugfix - skills loading\r\n\r\n### v0.9.8.1 - Caching, context optimization, error handling...\r\n- improved caching for anthropic models\r\n- improved context window optimimization\r\n- fixed retry on critical error counter\r\n- load up to 5 skills at a time\r\n- default chat model is now Sonnet 4.6\r\n\r\n\r\n### v0.9.8 - Skills, UI Redesign & Git projects\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FNV7s78yn6DY)\r\n\r\n- Skills\r\n    - Skills System replacing the legacy Instruments with a new `SKILL.md` standard for structured, portable agent capabilities.\r\n    - Built-in skills, and UI support for importing and listing skills\r\n- Real-time WebSocket infrastructure replacing the polling-based approach for UI state synchronization\r\n- UI Redesign\r\n    - Process groups to visually group agent actions with expand\u002Fcollapse support\r\n    - Timestamps, steps count and execution time with tool-specific badges\r\n    - Step detail modals with key-value and raw JSON display\r\n    - Collapsible responses with show more\u002Fless and copy buttons on code blocks and tables\r\n    - Message queue system allowing users to queue messages while the agent is still processing\r\n    - In-browser file editor for viewing and editing files without leaving the UI\r\n    - Welcome screen redesign with info and warning banners for connection security, missing API keys, and system resources\r\n    - Scheduler redesign with standalone modal, separate task list, detail and editor components, and project support\r\n    - Smooth response rendering and scroll stabilization across chat, terminals, and image viewer\r\n    - Chat width setting and reworked preferences panel\r\n    - Image viewer improvements with scroll support and expanded viewer\r\n    - Redesigned sidebar with reusable dropdown component and streamlined buttons\r\n    - Inline button confirmations for critical actions\r\n    - Improved login design and new logout button\r\n    - File browser enhanced with rename and file actions dropdown\r\n- Git projects\r\n    - Git-based projects with clone authentication for public and private repositories\r\n- Four new LLM providers: CometAPI, Z.AI, Moonshot AI, and AWS Bedrock\r\n- Microsoft Dev Tunnels integration for secure remote access\r\n- User data migration to `\u002Fusr` directory for cleaner separation of user and system files\r\n- Subagents system with configurable agent profiles for different roles\r\n- Memory operations offloaded to deferred tasks for better performance\r\n- Environment variables can now configure settings via `A0_SET_*` prefix in `.env`\r\n- Automatic migration with overwrite support for `.env`, scheduler, knowledge, and legacy directories\r\n- Projects support extended to MCP, A2A, and external API\r\n- Workdir outside project support for more flexible file organization\r\n- Agent number tracking in backend and responses for multi-agent identification\r\n- Many bug fixes and stability improvements across the UI, MCP tools, scheduler, uploads, and WebSocket handling\r\n","2026-03-24T20:46:34",{"id":200,"version":201,"summary_zh":202,"released_at":203},104751,"v0.9.8.2","### v0.9.8.2 - Bugfix - skills loading\r\n\r\n### v0.9.8.1 - Caching, context optimization, error handling...\r\n- improved caching for anthropic models\r\n- improved context window optimimization\r\n- fixed retry on critical error counter\r\n- load up to 5 skills at a time\r\n- default chat model is now Sonnet 4.6\r\n\r\n\r\n### v0.9.8 - Skills, UI Redesign & Git projects\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FNV7s78yn6DY)\r\n\r\n- Skills\r\n    - Skills System replacing the legacy Instruments with a new `SKILL.md` standard for structured, portable agent capabilities.\r\n    - Built-in skills, and UI support for importing and listing skills\r\n- Real-time WebSocket infrastructure replacing the polling-based approach for UI state synchronization\r\n- UI Redesign\r\n    - Process groups to visually group agent actions with expand\u002Fcollapse support\r\n    - Timestamps, steps count and execution time with tool-specific badges\r\n    - Step detail modals with key-value and raw JSON display\r\n    - Collapsible responses with show more\u002Fless and copy buttons on code blocks and tables\r\n    - Message queue system allowing users to queue messages while the agent is still processing\r\n    - In-browser file editor for viewing and editing files without leaving the UI\r\n    - Welcome screen redesign with info and warning banners for connection security, missing API keys, and system resources\r\n    - Scheduler redesign with standalone modal, separate task list, detail and editor components, and project support\r\n    - Smooth response rendering and scroll stabilization across chat, terminals, and image viewer\r\n    - Chat width setting and reworked preferences panel\r\n    - Image viewer improvements with scroll support and expanded viewer\r\n    - Redesigned sidebar with reusable dropdown component and streamlined buttons\r\n    - Inline button confirmations for critical actions\r\n    - Improved login design and new logout button\r\n    - File browser enhanced with rename and file actions dropdown\r\n- Git projects\r\n    - Git-based projects with clone authentication for public and private repositories\r\n- Four new LLM providers: CometAPI, Z.AI, Moonshot AI, and AWS Bedrock\r\n- Microsoft Dev Tunnels integration for secure remote access\r\n- User data migration to `\u002Fusr` directory for cleaner separation of user and system files\r\n- Subagents system with configurable agent profiles for different roles\r\n- Memory operations offloaded to deferred tasks for better performance\r\n- Environment variables can now configure settings via `A0_SET_*` prefix in `.env`\r\n- Automatic migration with overwrite support for `.env`, scheduler, knowledge, and legacy directories\r\n- Projects support extended to MCP, A2A, and external API\r\n- Workdir outside project support for more flexible file organization\r\n- Agent number tracking in backend and responses for multi-agent identification\r\n- Many bug fixes and stability improvements across the UI, MCP tools, scheduler, uploads, and WebSocket handling\r\n","2026-02-24T13:33:50",{"id":205,"version":206,"summary_zh":207,"released_at":208},104752,"v0.9.8.1","### v0.9.8.1 - Caching, context optimization, error handling...\r\n- improved caching for anthropic models\r\n- improved context window optimimization\r\n- fixed retry on critical error counter\r\n- load up to 5 skills at a time\r\n- default chat model is now Sonnet 4.6\r\n\r\n\r\n### v0.9.8 - Skills, UI Redesign & Git projects\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FNV7s78yn6DY)\r\n\r\n- Skills\r\n    - Skills System replacing the legacy Instruments with a new `SKILL.md` standard for structured, portable agent capabilities.\r\n    - Built-in skills, and UI support for importing and listing skills\r\n- Real-time WebSocket infrastructure replacing the polling-based approach for UI state synchronization\r\n- UI Redesign\r\n    - Process groups to visually group agent actions with expand\u002Fcollapse support\r\n    - Timestamps, steps count and execution time with tool-specific badges\r\n    - Step detail modals with key-value and raw JSON display\r\n    - Collapsible responses with show more\u002Fless and copy buttons on code blocks and tables\r\n    - Message queue system allowing users to queue messages while the agent is still processing\r\n    - In-browser file editor for viewing and editing files without leaving the UI\r\n    - Welcome screen redesign with info and warning banners for connection security, missing API keys, and system resources\r\n    - Scheduler redesign with standalone modal, separate task list, detail and editor components, and project support\r\n    - Smooth response rendering and scroll stabilization across chat, terminals, and image viewer\r\n    - Chat width setting and reworked preferences panel\r\n    - Image viewer improvements with scroll support and expanded viewer\r\n    - Redesigned sidebar with reusable dropdown component and streamlined buttons\r\n    - Inline button confirmations for critical actions\r\n    - Improved login design and new logout button\r\n    - File browser enhanced with rename and file actions dropdown\r\n- Git projects\r\n    - Git-based projects with clone authentication for public and private repositories\r\n- Four new LLM providers: CometAPI, Z.AI, Moonshot AI, and AWS Bedrock\r\n- Microsoft Dev Tunnels integration for secure remote access\r\n- User data migration to `\u002Fusr` directory for cleaner separation of user and system files\r\n- Subagents system with configurable agent profiles for different roles\r\n- Memory operations offloaded to deferred tasks for better performance\r\n- Environment variables can now configure settings via `A0_SET_*` prefix in `.env`\r\n- Automatic migration with overwrite support for `.env`, scheduler, knowledge, and legacy directories\r\n- Projects support extended to MCP, A2A, and external API\r\n- Workdir outside project support for more flexible file organization\r\n- Agent number tracking in backend and responses for multi-agent identification\r\n- Many bug fixes and stability improvements across the UI, MCP tools, scheduler, uploads, and WebSocket handling\r\n","2026-02-18T17:04:53",{"id":210,"version":211,"summary_zh":212,"released_at":213},104753,"v0.9.8","### v0.9.8 - Skills, UI Redesign & Git projects\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FNV7s78yn6DY)\r\n\r\n- Skills\r\n    - Skills System replacing the legacy Instruments with a new `SKILL.md` standard for structured, portable agent capabilities.\r\n    - Built-in skills, and UI support for importing and listing skills\r\n- Real-time WebSocket infrastructure replacing the polling-based approach for UI state synchronization\r\n- UI Redesign\r\n    - Process groups to visually group agent actions with expand\u002Fcollapse support\r\n    - Timestamps, steps count and execution time with tool-specific badges\r\n    - Step detail modals with key-value and raw JSON display\r\n    - Collapsible responses with show more\u002Fless and copy buttons on code blocks and tables\r\n    - Message queue system allowing users to queue messages while the agent is still processing\r\n    - In-browser file editor for viewing and editing files without leaving the UI\r\n    - Welcome screen redesign with info and warning banners for connection security, missing API keys, and system resources\r\n    - Scheduler redesign with standalone modal, separate task list, detail and editor components, and project support\r\n    - Smooth response rendering and scroll stabilization across chat, terminals, and image viewer\r\n    - Chat width setting and reworked preferences panel\r\n    - Image viewer improvements with scroll support and expanded viewer\r\n    - Redesigned sidebar with reusable dropdown component and streamlined buttons\r\n    - Inline button confirmations for critical actions\r\n    - Improved login design and new logout button\r\n    - File browser enhanced with rename and file actions dropdown\r\n- Git projects\r\n    - Git-based projects with clone authentication for public and private repositories\r\n- Four new LLM providers: CometAPI, Z.AI, Moonshot AI, and AWS Bedrock\r\n- Microsoft Dev Tunnels integration for secure remote access\r\n- User data migration to `\u002Fusr` directory for cleaner separation of user and system files\r\n- Subagents system with configurable agent profiles for different roles\r\n- Memory operations offloaded to deferred tasks for better performance\r\n- Environment variables can now configure settings via `A0_SET_*` prefix in `.env`\r\n- Automatic migration with overwrite support for `.env`, scheduler, knowledge, and legacy directories\r\n- Projects support extended to MCP, A2A, and external API\r\n- Workdir outside project support for more flexible file organization\r\n- Agent number tracking in backend and responses for multi-agent identification\r\n- Many bug fixes and stability improvements across the UI, MCP tools, scheduler, uploads, and WebSocket handling\r\n","2026-02-10T18:25:22",{"id":215,"version":216,"summary_zh":217,"released_at":218},104754,"v0.9.7","### v0.9.7 - Projects\r\n\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FRrTDp_v9V1c)\r\n\r\n- Projects management\r\n    - Support for custom instructions\r\n    - Integration with memory, knowledge, files\r\n    - Project specific secrets \r\n- New Welcome screen\u002FDashboard\r\n- New Wait tool\r\n- Subordinate agent configuration override support\r\n- Support for multiple documents at once in document_query_tool\r\n- Improved context on interventions\r\n- Openrouter embedding support\r\n- Frontend components refactor and polishing\r\n- SSH metadata output fix\r\n- Support for windows powershell in local TTY utility\r\n- More efficient selective streaming for LLMs\r\n- UI output length limit improvements\r\n- Update checker","2025-11-19T11:41:50",{"id":220,"version":221,"summary_zh":222,"released_at":223},104755,"v0.9.6","# v0.9.6 - Memory Dashboard\r\n[Release video](https:\u002F\u002Fyoutu.be\u002FsizjAq2-d9s)\r\n- Memory Management Dashboard\r\n- Kali update\r\n- Python update + dual installation\r\n- Browser Use update\r\n- New login screen\r\n- LiteLLM retry on temporary errors\r\n- Github Copilot provider support","2025-10-02T14:05:11",{"id":225,"version":226,"summary_zh":227,"released_at":228},104756,"v0.9.5.1","## 0.9.5.1\r\n- added support for Agent Zero Venice provider\r\n- added support for xAI provider\r\n\r\n## 0.9.5\r\n- Secrets management - agent can use credentials without seeing them\r\n- Agent can copy paste messages and files without rewriting them\r\n- LiteLLM global configuration field\r\n- Custom HTTP headers field for browser agent\r\n- Progressive web app support\r\n- Extra model params support for JSON\r\n- Short IDs for files and memories to prevent LLM errors\r\n- Tunnel component frontend rework\r\n- Fix for timezone change bug\r\n- Notifications z-index fix","2025-09-19T07:16:34",{"id":230,"version":231,"summary_zh":232,"released_at":233},104757,"v0.9.5","- Secrets management - agent can use credentials without seeing them\r\n- Agent can copy paste messages and files without rewriting them\r\n- LiteLLM global configuration field\r\n- Custom HTTP headers field for browser agent\r\n- Progressive web app support\r\n- Extra model params support for JSON\r\n- Short IDs for files and memories to prevent LLM errors\r\n- Tunnel component frontend rework\r\n- Fix for timezone change bug\r\n- Notifications z-index fix","2025-09-03T08:57:25",{"id":235,"version":236,"summary_zh":237,"released_at":238},104758,"v0.9.4","# v0.9.4 - Connectivity, UI\r\n- External API endpoints\r\n- Streamable HTTP MCP A0 server\r\n- A2A (Agent to Agent) protocol - server+client\r\n- New notifications system\r\n- New local terminal interface for stability\r\n- Rate limiter integration to models\r\n- Delayed memory recall\r\n- Smarter autoscrolling in UI\r\n- Action buttons in messages\r\n- Multiple API keys support\r\n- Tunnel URL QR code\r\n- Download streaming\r\n- Internal fixes and optimizations","2025-08-18T11:09:48",{"id":240,"version":241,"summary_zh":242,"released_at":243},104759,"v0.9.3","- Faster startup\u002Frestart\r\n- Subordinate agents can have dedicated prompts, tools and system extensions\r\n- Streamable HTTP MCP server support\r\n- Memory loading enhanced by AI filter\r\n- Memory AI consolidation when saving memories\r\n- Auto memory system configuration in settings\r\n- LLM providers available are set by providers.yaml configuration file\r\n- Venice.ai LLM provider supported\r\n- Initial agent message for user + as example for LLM\r\n- Docker build support for local images\r\n- File browser fix","2025-08-04T17:45:44",{"id":245,"version":246,"summary_zh":247,"released_at":248},104760,"v0.9.2","# v0.9.2 - Kokoro TTS, Attachments\r\n\r\n- Kokoro text-to-speech integration\r\n- New message attachments system\r\n- Minor updates: log truncation, hyperlink targets, component examples, api cleanup","2025-07-21T08:44:41",{"id":250,"version":251,"summary_zh":252,"released_at":253},104761,"v0.9.1.1","# v0.9.1.1\r\n- LLM API base URL field fix\r\n\r\n# v0.9.1 - LiteLLM, UI improvements\r\n- Langchain replaced with LiteLLM\r\n    - Support for reasoning models streaming\r\n    - Support for more providers\r\n    - Openrouter set as default instead of OpenAI\r\n- UI improvements\r\n    - New message grouping system\r\n    - Communication smoother and more efficient\r\n    - Collapsible messages by type\r\n    - Code execution tool output improved\r\n    - Tables and code blocks scrollable\r\n    - More space efficient on mobile\r\n- Streamable HTTP MCP servers support\r\n- LLM API URL added to models config for Azure, local and custom providers","2025-07-09T08:46:40",{"id":255,"version":256,"summary_zh":257,"released_at":258},104762,"v0.9.1","# v0.9.1 - LiteLLM, UI improvements\r\n- Langchain replaced with LiteLLM\r\n    - Support for reasoning models streaming\r\n    - Support for more providers\r\n    - Openrouter set as default instead of OpenAI\r\n- UI improvements\r\n    - New message grouping system\r\n    - Communication smoother and more efficient\r\n    - Collapsible messages by type\r\n    - Code execution tool output improved\r\n    - Tables and code blocks scrollable\r\n    - More space efficient on mobile\r\n- Streamable HTTP MCP servers support\r\n- LLM API URL added to models config for Azure, local and custom providers","2025-07-07T15:03:34"]