[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-agentscope-ai--CoPaw":3,"tool-agentscope-ai--CoPaw":61},[4,18,26,36,44,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},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,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},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",154349,2,"2026-04-13T23:32:16",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":32,"last_commit_at":42,"category_tags":43,"status":17},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。",108322,"2026-04-10T11:39:34",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":32,"last_commit_at":50,"category_tags":51,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[52,13,15,14],"插件",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":10,"last_commit_at":59,"category_tags":60,"status":17},4487,"LLMs-from-scratch","rasbt\u002FLLMs-from-scratch","LLMs-from-scratch 是一个基于 PyTorch 的开源教育项目，旨在引导用户从零开始一步步构建一个类似 ChatGPT 的大型语言模型（LLM）。它不仅是同名技术著作的官方代码库，更提供了一套完整的实践方案，涵盖模型开发、预训练及微调的全过程。\n\n该项目主要解决了大模型领域“黑盒化”的学习痛点。许多开发者虽能调用现成模型，却难以深入理解其内部架构与训练机制。通过亲手编写每一行核心代码，用户能够透彻掌握 Transformer 架构、注意力机制等关键原理，从而真正理解大模型是如何“思考”的。此外，项目还包含了加载大型预训练权重进行微调的代码，帮助用户将理论知识延伸至实际应用。\n\nLLMs-from-scratch 特别适合希望深入底层原理的 AI 开发者、研究人员以及计算机专业的学生。对于不满足于仅使用 API，而是渴望探究模型构建细节的技术人员而言，这是极佳的学习资源。其独特的技术亮点在于“循序渐进”的教学设计：将复杂的系统工程拆解为清晰的步骤，配合详细的图表与示例，让构建一个虽小但功能完备的大模型变得触手可及。无论你是想夯实理论基础，还是为未来研发更大规模的模型做准备",90106,"2026-04-06T11:19:32",[35,15,13,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":66,"readme_en":67,"readme_zh":68,"quickstart_zh":69,"use_case_zh":70,"hero_image_url":71,"owner_login":72,"owner_name":73,"owner_avatar_url":74,"owner_bio":75,"owner_company":76,"owner_location":76,"owner_email":77,"owner_twitter":76,"owner_website":76,"owner_url":78,"languages":79,"stars":117,"forks":118,"last_commit_at":119,"license":120,"difficulty_score":32,"env_os":121,"env_gpu":122,"env_ram":123,"env_deps":124,"category_tags":130,"github_topics":131,"view_count":32,"oss_zip_url":76,"oss_zip_packed_at":76,"status":17,"created_at":138,"updated_at":139,"faqs":140,"releases":171},7386,"agentscope-ai\u002FCoPaw","CoPaw","Your Personal AI Assistant; easy to install, deploy on your own machine or on the cloud; supports multiple chat apps with easily extensible capabilities.","QwenPaw 是一款专为个人打造的开源 AI 助手，旨在让用户轻松拥有完全自主控制的智能伙伴。它解决了用户在使用云端 AI 服务时普遍担忧的数据隐私与依赖问题，支持一键部署在本地电脑或私有云上，确保所有记忆与个性化数据仅留存于用户自己的设备中，无需经过第三方托管。\n\n除了基础对话，QwenPaw 具备强大的技能扩展能力，内置了日程管理、文档处理（PDF\u002FOffice）及新闻摘要等实用功能，并允许用户通过自定义插件无限拓展其能力边界。更独特的是，它支持多智能体协作模式，用户可以创建多个独立运作的 Agent 分工处理复杂任务，实现更高效的自动化流程。\n\n这款工具非常适合注重数据隐私的普通用户、希望搭建私有化 AI 服务的开发者，以及需要定制化工作流的研究人员。无论是想在本地离线运行以保障安全，还是希望在云端灵活调度，QwenPaw 都提供了极简的安装体验与开放的架构，真正实现了“为你所用，随你成长”的理念。","\u003Cdiv align=\"center\">\n\n# QwenPaw\n\n[![GitHub Repo](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGitHub-Repo-black.svg?logo=github)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fqwenpaw?color=3775A9&label=PyPI&logo=pypi)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fqwenpaw\u002F)\n[![Documentation](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocs-Website-green.svg?logo=readthedocs&label=Docs)](https:\u002F\u002Fqwenpaw.agentscope.io\u002F)\n[![Python Version](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10%20~%20%3C3.14-blue.svg?logo=python&label=Python)](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F)\n[![Last Commit](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002Fagentscope-ai\u002FQwenPaw)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-red.svg?logo=apache&label=License)](LICENSE)\n[![Code Style](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-black.svg?logo=python&label=CodeStyle)](https:\u002F\u002Fgithub.com\u002Fpsf\u002Fblack)\n[![GitHub Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=yellow&label=Stars)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fstargazers)\n[![GitHub Forks](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=purple&label=Forks)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fnetwork)\n[![DeepWiki](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeepWiki-Ask_Devin-navy.svg?logo=data:image\u002Fpng;base64,iVBORw0KGgoAAAANSUhEUgAAACwAAAAyCAYAAAAnWDnqAAAAAXNSR0IArs4c6QAAA05JREFUaEPtmUtyEzEQhtWTQyQLHNak2AB7ZnyXZMEjXMGeK\u002FAIi+QuHrMnbChYY7MIh8g01fJoopFb0uhhEqqcbWTp06\u002Fuv1saEDv4O3n3dV60RfP947Mm9\u002FSQc0ICFQgzfc4CYZoTPAswgSJCCUJUnAAoRHOAUOcATwbmVLWdGoH\u002F\u002FPB8mnKqScAhsD0kYP3j\u002FYt5LPQe2KvcXmGvRHcDnpxfL2zOYJ1mFwrryWTz0advv1Ut4CJgf5uhDuDj5eUcAUoahrdY\u002F56ebRWeraTjMt\u002F00Sh3UDtjgHtQNHwcRGOC98BJEAEymycmYcWwOprTgcB6VZ5JK5TAJ+fXGLBm3FDAmn6oPPjR4rKCAoJCal2eAiQp2x0vxTPB3ALO2CRkwmDy5WohzBDwSEFKRwPbknEggCPB\u002FimwrycgxX2NzoMCHhPkDwqYMr9tRcP5qNrMZHkVnOjRMWwLCcr8ohBVb1OMjxLwGCvjTikrsBOiA6fNyCrm8V1rP93iVPpwaE+gO0SsWmPiXB+jikdf6SizrT5qKasx5j8ABbHpFTx+vFXp9EnYQmLx02h1QTTrl6eDqxLnGjporxl3NL3agEvXdT0WmEost648sQOYAeJS9Q7bfUVoMGnjo4AZdUMQku50McDcMWcBPvr0SzbTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTogQkiqIhJV0nBQBTU+3okKCFDy9WwferkHjtxib7t3xIUQtHxnIwtx4mpg26\u002FHfwVNVDb4oI9RHmx5WGelRVlrtiw43zboCLaxv46AZeB3IlTkwouebTr1y2NjSpHz68WNFjHvupy3q8TFn3Hos2IAk4Ju5dCo8B3wP7VPr\u002FFGaKiG+T+v+TQqIrOqMTL1VdWV1DdmcbO8KXBz6esmYWYKPwDL5b5FA1a0hwapHiom0r\u002FcKaoqr+27\u002FXcrS5UwSMbQAAAABJRU5ErkJggg==)](https:\u002F\u002Fdeepwiki.com\u002Fagentscope-ai\u002FQwenPaw)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join_Us-blueviolet.svg?logo=discord)](https:\u002F\u002Fdiscord.gg\u002FeYMpfnkG8h)\n[![X](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-Follow_Us-black.svg?logo=x)](https:\u002F\u002Fx.com\u002Fagentscope_ai)\n[![DingTalk](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDingTalk-Join_Us-orange.svg)](https:\u002F\u002Fqr.dingtalk.com\u002Faction\u002Fjoingroup?code=v1,k1,OmDlBXpjW+I2vWjKDsjvI9dhcXjGZi3bQiojOq3dlDw=&_dt_no_comment=1&origin=11)\n\n[[Documentation](https:\u002F\u002Fqwenpaw.agentscope.io\u002F)] [[中文](README_zh.md)] [[日本語](README_ja.md)] [[Русский](README_ru.md)]\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgw.alicdn.com\u002Fimgextra\u002Fi1\u002FO1CN01sens5C1TuwioeGexL_!!6000000002443-55-tps-771-132.svg\" alt=\"QwenPaw Logo\" width=\"120\">\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cb>Works for you, grows with you.\u003C\u002Fb>\u003C\u002Fp>\n\n\u003C\u002Fdiv>\n\nYour personal AI assistant — easy to install, deploy locally or in the cloud, connect across channels, extend with ease.\n\n> **Core capabilities:**\n>\n> **Under your control** — Memory and personalization fully under your control. Deploy locally (data stays on your machine) or in the cloud (your chosen server). No third-party hosting, no data upload.\n>\n> **Skills extension** — Built-in scheduling, PDF\u002FOffice processing, news digest, and more; custom skills auto-loaded, no lock-in. Skills determine what QwenPaw can do.\n>\n> **Multi-agent collaboration** — Create multiple independent agents, each with their own role; enable collaboration skills for inter-agent communication to tackle complex tasks together.\n>\n> **Multi-layer security** — Tool guard, file access control, skill security scanning to ensure safe operation.\n>\n> **Every channel** — DingTalk, Feishu, WeChat, Discord, Telegram, and more. One QwenPaw, connect as needed.\n>\n> \u003Cdetails>\n> \u003Csummary>\u003Cb>What you can do with QwenPaw\u003C\u002Fb>\u003C\u002Fsummary>\n>\n> \u003Cbr>\n>\n> - **Social media**: Daily hot post digests (Xiaohongshu, Zhihu, Reddit), Bilibili\u002FYouTube video summaries.\n> - **Productivity**: Email & newsletter highlights pushed to DingTalk\u002FFeishu\u002FQQ; email & calendar contact organization.\n> - **Creative & building**: Describe your goal before sleep, auto-execute, wake up to a prototype; full workflow from topic selection to final video.\n> - **Research & learning**: Track tech & AI news, personal knowledge base search and reuse.\n> - **Desktop & files**: Organize and search local files, read & summarize documents, request files in chat.\n> - **Explore more**: Combine Skills with scheduled tasks into your own agentic app.\n>\n> \u003C\u002Fdetails>\n\n---\n\n## News\n\n[2026-04-12] **CoPaw is Officially Rebranding to QwenPaw**: This rebranding marks an important step forward into our next phase of open-source development.\n\nThe new name better reflects the open ecosystem we are building and the broader direction we are continuing to pursue:\n\n*Qwen* — represents a deeper integration with the Qwen open-source ecosystem and a sharper focus on the model layer, including local models for agents and smart collaboration between small and large models.\n\n*Paw* — carries forward our original mission to accompany users and become a personal assistant users can truly trust.\n\nThis rebranding does not change our vision and goals in the open-source community:\n- To build a more practical, secure, and personalized AI assistant;\n- To stay committed to open-source collaboration and move forward together with the community.\n\nThank you all for your support along the way.\nFrom today on, QwenPaw will stay and grow with you!\n\n---\n\n## Table of Contents\n\n> **Recommended reading:**\n>\n> - **🚀 I'm new and want to try quickly**: [Quick Start](#quick-start) → three commands to run → [Configure model](#api-key) → chat in Console\n> - **💬 I want to use in DingTalk\u002FFeishu\u002FWeChat**: Complete Quick Start → [Configure model](#api-key) → [Channel setup](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels)\n> - **🐍 I don't want to install Python**: [Desktop app](#option-6-desktop-application-beta) or [Script install](#option-2-script-install) or [ModelScope Studio](https:\u002F\u002Fmodelscope.cn\u002Fstudios\u002Ffork?target=AgentScope\u002FQwenPaw)\n> - **💻 I want to use local models (no API key needed)**: [Local Models](#local-models) → download model → start using\n> - **🛠️ I want to contribute code or develop new features**: [Install from source](#install-from-source) → [Contributing](#contributing)\n- [News](#news)\n- [Quick Start](#quick-start)\n- [API Key](#api-key)\n- [Local Models](#local-models)\n- [Documentation](#documentation)\n- [Security Features](#security-features)\n- [FAQ](#faq)\n- [Staying ahead](#staying-ahead)\n- [Roadmap](#roadmap)\n- [Install from source](#install-from-source)\n- [Contributing](#contributing)\n- [Why QwenPaw?](#why-qwenpaw)\n- [Built by](#built-by)\n- [License](#license)\n\n---\n\n## Quick Start\n\n### Option 1: pip install\n\nIf you prefer managing Python yourself:\n\n```bash\npip install qwenpaw\nqwenpaw init --defaults\nqwenpaw app\n```\n\nThen open the Console in your browser at **http:\u002F\u002F127.0.0.1:8088\u002F** to configure your model. To chat in DingTalk, Feishu, WeChat, etc., see the [Channel setup](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels) documentation.\n\n![Console](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_db5d3c5f9ec2.png)\n\n---\n\n### Option 2: Script install\n\nNo Python setup required, one command installs everything. The script will automatically download uv (Python package manager), create a virtual environment, and install QwenPaw with all dependencies (including Node.js and frontend assets). Note: May not work in restricted network environments or corporate firewalls.\n\n**macOS \u002F Linux:**\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash\n```\n\nTo install with Ollama support:\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash -s -- --extras ollama\n```\n\nTo install with multiple extras (e.g., Ollama + local):\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash -s -- --extras ollama,local\n```\n\n**Windows (CMD):**\n\n```CMD\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.bat -o install.bat && install.bat\n```\n\n**Windows (PowerShell):**\n\n```powershell\nirm https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.ps1 | iex\n```\n\n> **Note**: The installer will automatically check the status of uv. If it is not installed, it will attempt to download and configure it automatically. If the automatic installation fails, please follow the on-screen prompts or execute `python -m pip install -U uv`, then rerun the installer.\n\n> **⚠️ Special Notice for Windows Enterprise LTSC Users**\n>\n> If you are using Windows LTSC or an enterprise environment governed by strict security policies, PowerShell may run in **Constrained Language Mode**, potentially causing the following issue:\n> 1. **If using CMD (.bat): Script executes successfully but fails to write to `Path`**\n>\n>    The script completes file installation. Due to **Constrained Language Mode**, it cannot automatically update environment variables. Manually configure as follows:\n>    - **Locate the installation directory**:\n>      - Check if `uv` is available: Enter `uv --version` in CMD. If a version number appears, **only configure the QwenPaw path**. If you receive the prompt `'uv' is not recognized as an internal or external command, operable program or batch file,` configure both paths.\n>      - uv path (choose one based on installation location; use if `uv` fails): Typically `%USERPROFILE%\\.local\\bin`, `%USERPROFILE%\\AppData\\Local\\uv`, or the `Scripts` folder within your Python installation directory\n>      - QwenPaw path: Typically located at `%USERPROFILE%\\.qwenpaw\\bin`.\n>    - **Manually add to the system's Path environment variable**:\n>      - Press `Win + R`, type `sysdm.cpl` and press Enter to open System Properties.\n>      - Click “Advanced” -> “Environment Variables”.\n>      - Under “System variables”, locate and select `Path`, then click “Edit”.\n>      - Click “New”, enter both directory paths sequentially, then click OK to save.\n> 2. **If using PowerShell (.ps1): Script execution interrupted**\n>\n>   Due to **Constrained Language Mode**, the script may fail to automatically download `uv`.\n>   - **Manually install uv**: Refer to the [GitHub Release](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv\u002Freleases) to download `uv.exe` and place it in `%USERPROFILE%\\.local\\bin` or `%USERPROFILE%\\AppData\\Local\\uv`; or ensure Python is installed and run `python -m pip install -U uv`.\n>   - **Configure `uv` environment variables**: Add the `uv` directory and `%USERPROFILE%\\.qwenpaw\\bin` to your system's `Path` variable.\n>   - **Re-run the installation**: Open a new terminal and execute the installation script again to complete the `QwenPaw` installation.\n>   - **Configure the `QwenPaw` environment variable**: Add `%USERPROFILE%\\.qwenpaw\\bin` to your system's `Path` variable.\n\nOnce installed, open a new terminal and run:\n\n```bash\nqwenpaw init --defaults   # or: qwenpaw init (interactive)\nqwenpaw app\n```\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Install options\u003C\u002Fb>\u003C\u002Fsummary>\n\n**macOS \u002F Linux:**\n\n```bash\n# Install a specific version\ncurl -fsSL ... | bash -s -- --version 1.1.0\n\n# Install from source (dev\u002Ftesting)\ncurl -fsSL ... | bash -s -- --from-source\n\n# Upgrade — just re-run the installer\ncurl -fsSL ... | bash\n\n# Uninstall\nqwenpaw uninstall          # keeps config and data\nqwenpaw uninstall --purge  # removes everything\n```\n\n**Windows (PowerShell):**\n\n```powershell\n# Install a specific version\nirm ... | iex; .\\install.ps1 -Version 0.0.2\n\n# Install from source (dev\u002Ftesting)\n.\\install.ps1 -FromSource\n\n# Upgrade — just re-run the installer\nirm ... | iex\n\n# Uninstall\nqwenpaw uninstall          # keeps config and data\nqwenpaw uninstall --purge  # removes everything\n```\n\n\u003C\u002Fdetails>\n\n\n---\n\n### Option 3: Docker\n\nImages are on **Docker Hub** (`agentscope\u002Fqwenpaw`). Image tags: `latest` (stable); `pre` (PyPI pre-release).\n\n```bash\ndocker pull agentscope\u002Fqwenpaw:latest\ndocker run -p 127.0.0.1:8088:8088 \\\n  -v qwenpaw-data:\u002Fapp\u002Fworking \\\n  -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n  agentscope\u002Fqwenpaw:latest\n```\n\nAlso available on Alibaba Cloud Container Registry (ACR) for users in China: `agentscope-registry.ap-southeast-1.cr.aliyuncs.com\u002Fagentscope\u002Fqwenpaw` (same tags).\n\nThen open **http:\u002F\u002F127.0.0.1:8088\u002F** for the Console. Config, memory, and skills are stored in the `qwenpaw-data` volume; model provider settings and API keys are in the `qwenpaw-secrets` volume. To pass API keys (e.g. `DASHSCOPE_API_KEY`), add `-e VAR=value` or `--env-file .env` to `docker run`.\n\n> **Connecting to Ollama or other services on the host machine**\n>\n> Inside a Docker container, `localhost` refers to the container itself, not your host machine. If you run Ollama (or other model services) on the host and want QwenPaw in Docker to reach them, use one of these approaches:\n>\n> **Option A** — Explicit host binding (all platforms):\n> ```bash\n> docker run -p 127.0.0.1:8088:8088 \\\n>   --add-host=host.docker.internal:host-gateway \\\n>   -v qwenpaw-data:\u002Fapp\u002Fworking \\\n>   -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n>   agentscope\u002Fqwenpaw:latest\n> ```\n> Then in QwenPaw **Settings → Models**, change the Base URL to `http:\u002F\u002Fhost.docker.internal:\u003Cport>` — for example, `http:\u002F\u002Fhost.docker.internal:11434` for Ollama, or `http:\u002F\u002Fhost.docker.internal:1234\u002Fv1` for LM Studio.\n>\n> **Option B** — Host networking (Linux only):\n> ```bash\n> docker run --network=host \\\n>   -v qwenpaw-data:\u002Fapp\u002Fworking \\\n>   -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n>   agentscope\u002Fqwenpaw:latest\n> ```\n> No port mapping (`-p`) is needed; the container shares the host network directly. Note that all container ports are exposed on the host, which may cause conflicts if the port is already in use.\n>\n> **Note:** If you only mount `\u002Fapp\u002Fworking` without a separate volume for `\u002Fapp\u002Fworking.secret`, the entrypoint will automatically redirect secrets into `\u002Fapp\u002Fworking\u002F.secret` so they persist on the same volume.\n\nThe image is built from scratch. To build the image yourself, please refer to the [Build Docker image](scripts\u002FREADME.md#build-docker-image) section in `scripts\u002FREADME.md`, and then push to your registry.\n\n---\n\n### Option 4: Deploy on Alibaba Cloud ECS\n\nTo run QwenPaw on Alibaba Cloud (ECS), use the one-click deployment: open the [QwenPaw on Alibaba Cloud (ECS) deployment link](https:\u002F\u002Fcomputenest.console.aliyun.com\u002Fservice\u002Finstance\u002Fcreate\u002Fcn-hangzhou?type=user&ServiceId=service-1ed84201799f40879884) and follow the prompts. For step-by-step instructions, see [Alibaba Cloud Developer: Deploy your AI assistant in 3 minutes](https:\u002F\u002Fdeveloper.aliyun.com\u002Farticle\u002F1713682).\n\n---\n\n### Option 5: Using ModelScope\n\n**No local install?** [ModelScope Studio](https:\u002F\u002Fmodelscope.cn\u002Fstudios\u002Ffork?target=AgentScope\u002FQwenPaw) one-click cloud setup. Set your Studio to **non-public** so others cannot control your QwenPaw.\n\n---\n\n### Option 6: Desktop Application (Beta)\n\n> **Beta Notice**: The desktop application is currently in Beta testing phase with the following known limitations:\n> - **Incomplete compatibility testing**: Not fully tested across all system versions and hardware configurations\n> - **Potential performance issues**: Startup time, memory usage, and other performance aspects may need further optimization\n> - **Features under development**: Some features may be unstable or missing\n\nIf you're not comfortable with command-line tools, you can download and use QwenPaw's desktop application without manually configuring Python environments or running commands.\n\n#### Download\n\nDownload the desktop app from [GitHub Releases](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Freleases):\n- **Windows**: `QwenPaw-Setup-\u003Cversion>.exe`\n- **macOS**: `QwenPaw-\u003Cversion>-macOS.zip` (Apple Silicon recommended)\n\n#### Features\n\n- ✅ **Zero configuration**: Download and double-click to run, no need to install Python or configure environment variables\n- ✅ **Cross-platform**: Supports Windows 10+ and macOS 14+\n- ✅ **Visual interface**: Automatically opens browser interface, no need to manually enter addresses\n- ⚠️ **Beta stage**: Features are continuously being improved, feedback welcome\n\n#### First Launch\n\n**Important**: The first launch may take 10-60 seconds (depending on your system configuration). The application needs to initialize the Python environment and load dependencies. Please wait patiently for the browser window to open automatically.\n\n#### macOS: Bypass System Security Restrictions\n\nWhen you download the QwenPaw macOS app from Releases, macOS may show: *\"Apple cannot verify that 'QwenPaw' contains no malicious software\"*. This happens because the app is not notarized. You can still open it as follows:\n\n- **Right-click to open (recommended)**\n  Right-click (or Control+click) the QwenPaw app → **Open** → in the dialog click **Open** again. This tells Gatekeeper you trust the app; after that you can double-click to launch as usual.\n\n- **Allow in System Settings**\n  If it is still blocked, go to **System Settings → Privacy & Security**, scroll to the message like *\"QwenPaw was blocked because it is from an unidentified developer\"*, and click **Open Anyway** or **Allow**.\n\n- **Remove quarantine attribute (not recommended for most users)**\n  In Terminal run:\n  `xattr -cr \u002FApplications\u002FQwenPaw.app`\n  (or use the path to the `.app` after unzipping). This clears the \"downloaded from the internet\" quarantine flag so the warning usually does not appear, but is less safe and controllable than using **Right-click → Open**.\n\nFor detailed usage instructions, troubleshooting, and common issues, see the [Desktop Application Guide](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fdesktop).\n\n---\n\n## API Key\n\nIf you use a **cloud LLM API** (e.g., Qianwen, Gemini, OpenAI), you must configure an API key before chatting. QwenPaw will not work until a valid key is set. See the [official docs](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels) for details.\n\n**How to configure:**\n\n1. **Console (recommended)** — After running `qwenpaw app`, open **http:\u002F\u002F127.0.0.1:8088\u002F** → **Settings** → **Models**. Choose a provider, enter the **API Key**, and enable that provider and model.\n2. **`qwenpaw init`** — When you run `qwenpaw init`, it will guide you through configuring the LLM provider and API key. Follow the prompts to choose a provider and enter your key.\n3. **Environment variable** — For DashScope you can set `DASHSCOPE_API_KEY` in your shell or in a `.env` file in the working directory.\n\nTools that need extra keys (e.g. `TAVILY_API_KEY` for web search) can be set in Console **Settings → Environment variables**, see [Config](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconfig) for details.\n\n> **Using local models only?** If you use [Local Models](#local-models) (llama.cpp \u002F Ollama \u002F LM Studio), you do **not** need any API key.\n\n## Local Models\n\nQwenPaw can run LLMs entirely on your machine — no API keys or cloud services required. See the [official docs](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels) for details.\n\n| Backend       | Best for                                 | Install                                                              |\n| ------------- | ---------------------------------------- | -------------------------------------------------------------------- |\n| **llama.cpp** | Cross-platform (macOS \u002F Linux \u002F Windows) | No extra installation required; click \"Download Llama.cpp\" in the web UI. |\n| **Ollama**    | Cross-platform (requires Ollama service) | Install and start the Ollama app in advance. |\n| **LM Studio** | Cross-platform (requires LM Studio service) | Install and start the LM Studio app in advance. |\n\n---\n\n## Documentation\n\n| Topic                                                                 | Description                                      |\n| --------------------------------------------------------------------- | ------------------------------------------------ |\n| [Introduction](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fintro)                | What QwenPaw is and how to use it                  |\n| [Quick start](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fquickstart)            | Install and run (local or ModelScope Studio)    |\n| [Console](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconsole)                   | Web UI: chat and agent configuration            |\n| [Models](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels)                     | Configure cloud, local, and custom providers    |\n| [Channels](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels)                  | DingTalk, Feishu, QQ, Discord, iMessage, and more |\n| [Skills](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fskills)                      | Extend and customize capabilities               |\n| [MCP](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmcp)                            | Manage MCP clients                               |\n| [Memory](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmemory)                     | Long-term memory                     |\n| [Context](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcontext)                   | Context management mechanism                     |\n| [Magic commands](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcommands)           | Control conversation state without waiting for the AI |\n| [Heartbeat](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fheartbeat)                | Scheduled check-in and digest                    |\n| [Multi-Agent](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmulti-agent)           | Create multiple agents and enable collaboration  |\n| [Config & working dir](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconfig) | Working directory and config file                |\n| [CLI](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcli)                            | Init, cron jobs, skills, clean                   |\n| [FAQ](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Ffaq)                           | Common questions and troubleshooting             |\n\nFull docs in this repo: [website\u002Fpublic\u002Fdocs\u002F](website\u002Fpublic\u002Fdocs\u002F).\n\n---\n\n## Security Features\n\nQwenPaw includes multi-layer security mechanisms to protect your data and system:\n\n- **Tool guard** — Automatically intercepts dangerous shell commands (e.g., `rm -rf \u002F`, fork bombs, reverse shells, etc.)\n- **File access guard** — Restricts agent access to sensitive paths (e.g., `~\u002F.ssh`, key files, system directories, etc.)\n- **Skill security scanning** — Automatically scans before installing skills, detecting risks like prompt injection, command injection, hardcoded keys, data exfiltration, etc.\n- **Local deployment** — All data and memory stored locally, no third-party upload (when using cloud LLM APIs, conversation content is sent to the corresponding API provider)\n- **Web Authentication** — Optional login protection for the Console. Disabled by default; set `QWENPAW_AUTH_ENABLED=true` to enable. See [Web Authentication](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fsecurity#Web-Authentication) for details.\n\nSee [Security documentation](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fsecurity) for details.\n\n---\n\n## FAQ\n\nFor common questions, troubleshooting tips, and known issues, please visit the **[FAQ page](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Ffaq)**.\n\n---\n\n## Staying ahead\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_19d1165bd63e.gif\" width=\"600\" alt=\"Star QwenPaw\" \u002F>\n\u003C\u002Fa>\n\nStar QwenPaw on GitHub and be instantly notified of new releases.\n\n---\n\n## Roadmap\n\n| Area                                  | Item                                                                                                                                             | Status               |\n| ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------- |\n| **Horizontal Expansion**              | More channels, models, skills, MCPs — **community contributions welcome**                                                                        | Seeking Contributors |\n| **Existing Feature Extension**        | Display optimization, download hints, Windows path compatibility, etc. — **community contributions welcome**                                     | Seeking Contributors |\n| **Console Web UI**                    | Expose more info\u002Fconfig in the Console                                                                                                           | In Progress          |\n| **Multi-agent**                      | Agentic Ralph Loop                                                                                  | In Progress          |\n| **Multimodal**                        | Voice\u002Fvideo calls and real-time interaction                                                                                                      | In Progress          |\n| **Small + Large Model Collaboration** | Multi-model routing; different models for different tasks                                                                                          | In Progress          |\n| **Memory System**                     | Experience distillation & skill extraction                                                                                                       | In Progress          |\n|                                       | Memory mechanism switching                                                                                                                       | In Progress             |\n|                                       | Multimodal memory fusion                                                                                                                         | Planned              |\n|                                       | Context-aware proactive delivery                                                                                                                 | Planned              |\n| **Sandbox**                           | Deeper integration with AgentScope Runtime sandboxes                                                                                             | In Progress          |\n| **Cloud-native**                      | Deeper integration with AgentScope Runtime; leverage cloud compute, storage, tools, and skills                                                         | In Progress              |\n| **Skills Hub**                        | Enrich the [AgentScope Skills](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope-skills) repository and improve discoverability of high-quality skills | Planned              |\n\n\n_Status:_ **In Progress** — actively being worked on; **Planned** — queued or under design, also welcome contributions; **Seeking Contributors** — we strongly encourage community contributions.\n\n## Install from source\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw.git\ncd QwenPaw\n\n# Build console frontend first (required for web UI)\ncd console && npm ci && npm run build\ncd ..\n\n# Copy console build output to package directory\nmkdir -p src\u002Fqwenpaw\u002Fconsole\ncp -R console\u002Fdist\u002F. src\u002Fqwenpaw\u002Fconsole\u002F\n\n# Install Python package\npip install -e .\n```\n\n- **Dev** (tests, formatting): `pip install -e \".[dev,full]\"`\n- **Then**: Run `qwenpaw init --defaults`, then `qwenpaw app`.\n\n> **Note for updates:** When updating to a new major version after `git pull`, please also rebuild the frontend, reinstall the package (`pip install -e .`), restart `qwenpaw app`, and clear your browser cache with `Ctrl+Shift+R` (or `Cmd+Shift+R` on macOS).\n\n---\n\n## Contributing\n\nQwenPaw evolves through open collaboration, and we welcome all forms of contribution! Check the [Roadmap](#roadmap) above (especially items marked **Seeking Contributors**) to find areas that interest you, and read [CONTRIBUTING](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fblob\u002Fmain\u002FCONTRIBUTING.md) to get started. We particularly welcome:\n\n- **Horizontal expansion** — new channels, model providers, skills, MCPs.\n- **Existing feature extension & refinement** — display and interaction improvements, download hints, Windows path compatibility, etc.\n\nJoin [GitHub Discussions](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fdiscussions) to discuss ideas or pick up tasks.\n\n---\n\n## Why QwenPaw?\n\nQwenPaw stands for Qwen Personal Agent Workstation, and also embodies the wisdom of Qwen and the warmth of a Paw.\nWe hope it is not a cold tool, but an intelligent and warm \"little paw\" always ready to help—a most intuitive partner in your digital life.\n\n---\n\n## Built by\n\n[AgentScope team](https:\u002F\u002Fgithub.com\u002Fagentscope-ai) · [AgentScope](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope) · [AgentScope Runtime](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope-runtime) · [ReMe](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FReMe)\n\n---\n\n## Contact us\n\n| [Discord](https:\u002F\u002Fdiscord.gg\u002FeYMpfnkG8h)                     | [X (Twitter)](https:\u002F\u002Fx.com\u002Fagentscope_ai)                   | [DingTalk](https:\u002F\u002Fqr.dingtalk.com\u002Faction\u002Fjoingroup?code=v1,k1,OmDlBXpjW+I2vWjKDsjvI9dhcXjGZi3bQiojOq3dlDw=&_dt_no_comment=1&origin=11) | [RedNote](https:\u002F\u002Fwww.xiaohongshu.com\u002Fuser\u002Fprofile\u002F691c18db0000000037032be9) |\n| ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |\n| [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_5e73a7413fa9.png\" width=\"80\" height=\"80\" alt=\"Discord\">](https:\u002F\u002Fdiscord.gg\u002FeYMpfnkG8h) | [\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-black.svg?logo=x&logoColor=white\" width=\"80\" height=\"80\" alt=\"X\">](https:\u002F\u002Fx.com\u002Fagentscope_ai) | [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_584abd3b9a38.png\" width=\"80\" height=\"80\" alt=\"DingTalk\">](https:\u002F\u002Fqr.dingtalk.com\u002Faction\u002Fjoingroup?code=v1,k1,OmDlBXpjW+I2vWjKDsjvI9dhcXjGZi3bQiojOq3dlDw=&_dt_no_comment=1&origin=11) | [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_1e0690be8700.jpg\" width=\"80\" height=\"80\" alt=\"RedNote\">](https:\u002F\u002Fwww.xiaohongshu.com\u002Fuser\u002Fprofile\u002F691c18db0000000037032be9) |\n\n---\n\n## Telemetry\n\nQwenPaw collects **anonymous** usage data during `qwenpaw init` to help us understand our user base and prioritize improvements. Data is sent **once per version** — when you upgrade QwenPaw, telemetry is re-collected so we can track version adoption.\n\n**What we collect:**\n\n- QwenPaw version (e.g., 0.0.7)\n- Install method (pip, Docker, or desktop app)\n- OS and version (e.g., macOS 14.0, Ubuntu 22.04)\n- Python version (e.g., 3.13)\n- CPU architecture (e.g., x86_64, arm64)\n- GPU availability (yes\u002Fno)\n\n**What we do NOT collect:** No personal data, no files, no credentials, no IP addresses, no identifiable information.\n\nWhen running `qwenpaw init` interactively, you will be asked whether to opt in. If you choose `--defaults`, telemetry is accepted automatically. The prompt appears once per version and never affects QwenPaw's functionality.\n\n---\n\n## License\n\nQwenPaw is released under the [Apache License 2.0](LICENSE).\n\n---\n\n## Contributors\n\nAll thanks to our contributors:\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_2751d8b17041.png\" alt=\"Contributors\" \u002F>\n\u003C\u002Fa>\n","\u003Cdiv align=\"center\">\n\n# QwenPaw\n\n[![GitHub Repo](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGitHub-Repo-black.svg?logo=github)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fqwenpaw?color=3775A9&label=PyPI&logo=pypi)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fqwenpaw\u002F)\n[![Documentation](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocs-Website-green.svg?logo=readthedocs&label=Docs)](https:\u002F\u002Fqwenpaw.agentscope.io\u002F)\n[![Python Version](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10%20~%20%3C3.14-blue.svg?logo=python&label=Python)](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F)\n[![Last Commit](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002Fagentscope-ai\u002FQwenPaw)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-red.svg?logo=apache&label=License)](LICENSE)\n[![Code Style](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-black.svg?logo=python&label=CodeStyle)](https:\u002F\u002Fgithub.com\u002Fpsf\u002Fblack)\n[![GitHub Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=yellow&label=Stars)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fstargazers)\n[![GitHub Forks](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=purple&label=Forks)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fnetwork)\n[![DeepWiki](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeepWiki-Ask_Devin-navy.svg?logo=data:image\u002Fpng;base64,iVBORw0KGgoAAAANSUhEUgAAACwAAAAyCAYAAAAnWDnqAAAAAXNSR0IArs4c6QAAA05JREFUaEPtmUtyEzEQhtWTQyQLHNak2AB7ZnyXZMEjXMGeK\u002FAIi+QuHrMnbChYY7MIh8g01fJoopFb0uhhEqqcbWTp06\u002Fuv1saEDv4O3n3dV60RfP947Mm9\u002FSQc0ICFQgzfc4CYZoTPAswgSJCCUJUnAAoRHOAUOcATwbmVLWdGoH\u002F\u002FPB8mnKqScAhsD0kYP3j\u002FYt5LPQe2KvcXmGvRHcDnpxfL2zOYJ1mFwrryWTz0advv1Ut4CJgf5uhDuDj5eUcAUoahrdY\u002F56ebRWeraTjMt\u002F00Sh3UDtjgHtQNHwcRGOC98BJEAEymycmYcWwOprTgcB6VZ5JK5TAJ+fXGLBm3FDAmn6oPPjR4rKCAoJCal2eAiQp2x0vxTPB3ALO2CRkwmDy5WohzBDwSEFKRwPbknEggCPB\u002FimwrycgxX2NzoMCHhPkDwqYMr9tRcP5qNrMZHkVnOjRMWwLCcr8ohBVb1OMjxLwGCvjTikrsBOiA6fNyCrm8V1rP93iVPpwaE+gO0SsWmPiXB+jikdf6SizrT5qKasx5j8ABbHpFTx+vFXp9EnYQmLx02h1QTTrl6eDqxLnGjporxl3NL3agEvXdT0WmEost648sQOYAeJS9Q7bfUVoMGnjo4AZdUMQku50McDcMWcBPvr0SzbTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wD......\u003Cdiv align=\"center\">\n\n# QwenPaw\n\n[![GitHub 仓库](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGitHub-Repo-black.svg?logo=github)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fqwenpaw?color=3775A9&label=PyPI&logo=pypi)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fqwenpaw\u002F)\n[![文档](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocs-Website-green.svg?logo=readthedocs&label=Docs)](https:\u002F\u002Fqwenpaw.agentscope.io\u002F)\n[![Python 版本](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10%20~%20%3C3.14-blue.svg?logo=python&label=Python)](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F)\n[![最近提交](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002Fagentscope-ai\u002FQwenPaw)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw)\n[![许可证](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-red.svg?logo=apache&label=License)](LICENSE)\n[![代码风格](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-black.svg?logo=python&label=CodeStyle)](https:\u002F\u002Fgithub.com\u002Fpsf\u002Fblack)\n[![GitHub 星标](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=yellow&label=Stars)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fstargazers)\n[![GitHub 分支](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002Fagentscope-ai\u002FQwenPaw?style=flat&logo=github&color=purple&label=Forks)](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fnetwork)\n[![DeepWiki](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeepWiki-Ask_Devin-navy.svg?logo=data:image\u002Fpng;base64,iVBORw0KGgoAAAANSUhEUgAAACwAAAAyCAYAAAAnWDnqAAAAAXNSR0IArs4c6QAAA05JREFUaEPtmUtyEzEQhtWTQyQLHNak2AB7ZnyXZMEjXMGeK\u002FAIi+QuHrMnbChYY7MIh8g01fJoopFb0uhhEqqcbWTp06\u002Fuv1saEDv4O3n3dV60RfP947Mm9\u002FSQc0ICFQgzfc4CYZoTPAswgSJCCUJUnAAoRHOAUOcATwbmVLWdGoH\u002F\u002FPB8mnKqScAhsD0kYP3j\u002FYt5LPQe2KvcXmGvRHcDnpxfL2zOYJ1mFwrryWTz0advv1Ut4CJgf5uhDuDj5eUcAUoahrdY\u002F56ebRWeraTjMt\u002F00Sh3UDtjgHtQNHwcRGOC98BJEAEymycmYcWwOprTgcB6VZ5JK5TAJ+fXGLBm3FDAmn6oPPjR4rKCAoJCal2eAiQp2x0vxTPB3ALO2CRkwmDy5WohzBDwSEFKRwPbknEggCPB\u002FimwrycgxX2NzoMCHhPkDwqYMr9tRcP5qNrMZHkVnOjRMWwLCcr8ohBVb1OMjxLwGCvjTikrsBOiA6fNyCrm8V1rP93iVPpwaE+gO0SsWmPiXB+jikdf6SizrT5qKasx5j8ABbHpFTx+vFXp9EnYQmLx02h1QTTrl6eDqxLnGjporxl3NL3agEvXdT0WmEost648sQOYAeJS9Q7bfUVoMGnjo4AZdUMQku50McDcMWcBPvr0SzbTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai42ppWPKiBagOvaYk8lO7DajerabOZP46Lby5wKjw1HCRx7p9sVMOWGzb\u002FvA1hwiWc6jm3MvQDTAFDfvJqwLzgxwATnCgnp4wDl6Aa+Ax283gghmj+vj7feE2KBBRMW3FzOpLOADl0Isb5587h\u002FU4gGvkt5v60Z1VLG8BhYjbzRwyQZemwAd6cCR5\u002FXFWLYZRIMpX39AR0tjaGGiGzLVyhse5C9RKC6ai......\n\n## 目录\n\n> **推荐阅读：**\n>\n> - **🚀 我是新手，想快速上手**：[快速入门](#quick-start) → 执行三条命令 → [配置模型](#api-key) → 在控制台聊天\n> - **💬 我想在钉钉\u002F飞书\u002F微信中使用**：完成快速入门 → [配置模型](#api-key) → [渠道设置](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels)\n> - **🐍 我不想安装 Python**：[桌面应用](#option-6-desktop-application-beta) 或 [脚本安装](#option-2-script-install) 或 [ModelScope Studio](https:\u002F\u002Fmodelscope.cn\u002Fstudios\u002Ffork?target=AgentScope\u002FQwenPaw)\n> - **💻 我想使用本地模型（无需 API 密钥）**：[本地模型](#local-models) → 下载模型 → 开始使用\n> - **🛠️ 我想贡献代码或开发新功能**：[从源码安装](#install-from-source) → [贡献指南](#contributing)\n- [新闻](#news)\n- [快速入门](#quick-start)\n- [API 密钥](#api-key)\n- [本地模型](#local-models)\n- [文档](#documentation)\n- [安全特性](#security-features)\n- [常见问题解答](#faq)\n- [保持领先](#staying-ahead)\n- [路线图](#roadmap)\n- [从源码安装](#install-from-source)\n- [贡献指南](#contributing)\n- [为什么选择 QwenPaw？](#why-qwenpaw)\n- [由谁构建](#built-by)\n- [许可证](#license)\n\n---\n\n## 快速入门\n\n### 选项 1：pip 安装\n\n如果您更喜欢自行管理 Python：\n\n```bash\npip install qwenpaw\nqwenpaw init --defaults\nqwenpaw app\n```\n\n然后在浏览器中打开控制台，地址为 **http:\u002F\u002F127.0.0.1:8088\u002F**，以配置您的模型。如需在钉钉、飞书、微信等平台聊天，请参阅 [渠道设置](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels) 文档。\n\n![控制台](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_db5d3c5f9ec2.png)\n\n---\n\n### 选项 2：脚本安装\n\n无需任何 Python 环境配置，一条命令即可完成所有安装。该脚本会自动下载 uv（Python 包管理器），创建虚拟环境，并安装 QwenPaw 及其所有依赖项（包括 Node.js 和前端资源）。注意：在受限网络环境或企业防火墙下可能无法正常工作。\n\n**macOS \u002F Linux：**\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash\n```\n\n若需支持 Ollama：\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash -s -- --extras ollama\n```\n\n若需同时支持多个扩展（例如 Ollama + 本地模型）：\n\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash -s -- --extras ollama,local\n```\n\n**Windows (CMD)：**\n\n```CMD\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.bat -o install.bat && install.bat\n```\n\n**Windows (PowerShell)：**\n\n```powershell\nirm https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.ps1 | iex\n```\n\n> **注意**：安装程序会自动检查 uv 的状态。如果未安装，它将尝试自动下载并配置。若自动安装失败，请按照屏幕提示操作，或执行 `python -m pip install -U uv`，然后重新运行安装程序。\n\n> **⚠️ 针对 Windows Enterprise LTSC 用户的特别说明**\n>\n> 如果您使用的是 Windows LTSC 或受严格安全策略约束的企业环境，PowerShell 可能会以 **受限语言模式** 运行，从而导致以下问题：\n> 1. **若使用 CMD (.bat)**：脚本成功执行，但无法写入 `Path`\n>\n>    脚本已完成文件安装。由于 **受限语言模式**，无法自动更新环境变量。请手动按如下步骤配置：\n>    - **找到安装目录**：\n>      - 检查是否已安装 `uv`：在 CMD 中输入 `uv --version`。若显示版本号，则只需配置 QwenPaw 的路径；若出现提示 `'uv' is not recognized as an internal or external command, operable program or batch file,`，则需同时配置两个路径。\n>      - uv 的路径（根据安装位置选择；若 `uv` 未成功安装，则使用此路径）：通常为 `%USERPROFILE%\\.local\\bin`、`%USERPROFILE%\\AppData\\Local\\uv` 或 Python 安装目录中的 `Scripts` 文件夹。\n>      - QwenPaw 的路径：通常位于 `%USERPROFILE%\\.qwenpaw\\bin`。\n>    - **手动添加到系统的 Path 环境变量**：\n>      - 按下 `Win + R`，输入 `sysdm.cpl` 并按 Enter 打开系统属性。\n>      - 点击“高级”→“环境变量”。\n>      - 在“系统变量”中，找到并选择 `Path`，然后点击“编辑”。\n>      - 点击“新建”，依次输入两个目录路径，然后点击确定保存。\n> 2. **若使用 PowerShell (.ps1)**：脚本执行中断\n>\n>   由于 **受限语言模式**，脚本可能无法自动下载 `uv`。\n>   - **手动安装 uv**：请参考 [GitHub 发布页面](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv\u002Freleases) 下载 `uv.exe`，并将其放置在 `%USERPROFILE%\\.local\\bin` 或 `%USERPROFILE%\\AppData\\Local\\uv`；或者确保已安装 Python，并运行 `python -m pip install -U uv`。\n>   - **配置 `uv` 的环境变量**：将 `uv` 目录和 `%USERPROFILE%\\.qwenpaw\\bin` 添加到系统的 `Path` 变量中。\n>   - **重新运行安装脚本**：打开一个新的终端，再次执行安装脚本，以完成 `QwenPaw` 的安装。\n>   - **配置 `QwenPaw` 的环境变量**：将 `%USERPROFILE%\\.qwenpaw\\bin` 添加到系统的 `Path` 变量中。\n\n安装完成后，打开一个新的终端并运行：\n\n```bash\nqwenpaw init --defaults   # 或：qwenpaw init（交互式）\nqwenpaw app\n```\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>安装选项\u003C\u002Fb>\u003C\u002Fsummary>\n\n**macOS \u002F Linux：**\n\n```bash\n# 安装特定版本\ncurl -fsSL ... | bash -s -- --version 1.1.0\n\n# 从源码安装（开发\u002F测试版）\ncurl -fsSL ... | bash -s -- --from-source\n\n# 升级——只需重新运行安装程序\ncurl -fsSL ... | bash\n\n# 卸载\nqwenpaw uninstall          # 保留配置和数据\nqwenpaw uninstall --purge  # 删除所有内容\n```\n\n**Windows (PowerShell)：**\n\n```powershell\n# 安装特定版本\nirm ... | iex; .\\install.ps1 -Version 0.0.2\n\n# 从源码安装（开发\u002F测试版）\n.\\install.ps1 -FromSource\n\n# 升级——只需重新运行安装程序\nirm ... | iex\n\n# 卸载\nqwenpaw uninstall          # 保留配置和数据\nqwenpaw uninstall --purge  # 删除所有内容\n```\n\n\u003C\u002Fdetails>\n\n\n---\n\n### 选项 3：Docker\n\n镜像托管在 **Docker Hub** 上（`agentscope\u002Fqwenpaw`）。镜像标签：`latest`（稳定版）；`pre`（PyPI 预发布版）。\n\n```bash\ndocker pull agentscope\u002Fqwenpaw:latest\ndocker run -p 127.0.0.1:8088:8088 \\\n  -v qwenpaw-data:\u002Fapp\u002Fworking \\\n  -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n  agentscope\u002Fqwenpaw:latest\n```\n\n同时，面向中国用户，该镜像也在阿里云容器镜像服务（ACR）上提供：`agentscope-registry.ap-southeast-1.cr.aliyuncs.com\u002Fagentscope\u002Fqwenpaw`（标签相同）。\n\n随后打开 **http:\u002F\u002F127.0.0.1:8088\u002F** 即可进入控制台。配置、内存和技能存储在 `qwenpaw-data` 卷中；模型提供商的设置和 API 密钥则存储在 `qwenpaw-secrets` 卷中。若需传递 API 密钥（如 `DASHSCOPE_API_KEY`），可在 `docker run` 命令中添加 `-e VAR=value` 或 `--env-file .env`。\n\n> **连接宿主机上的 Ollama 或其他服务**\n>\n> 在 Docker 容器内，`localhost` 指的是容器本身，而非宿主机。如果您在宿主机上运行 Ollama（或其他模型服务），并希望 Docker 中的 QwenPaw 能够访问它们，可以采用以下方法之一：\n>\n> **选项 A** — 显式绑定宿主机（适用于所有平台）：\n> ```bash\n> docker run -p 127.0.0.1:8088:8088 \\\n>   --add-host=host.docker.internal:host-gateway \\\n>   -v qwenpaw-data:\u002Fapp\u002Fworking \\\n>   -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n>   agentscope\u002Fqwenpaw:latest\n> ```\n> 然后在 QwenPaw 的 **设置 → 模型** 中，将基础 URL 更改为 `http:\u002F\u002Fhost.docker.internal:\u003C端口>` — 例如，对于 Ollama 可设置为 `http:\u002F\u002Fhost.docker.internal:11434`，对于 LM Studio 则为 `http:\u002F\u002Fhost.docker.internal:1234\u002Fv1`。\n>\n> **选项 B** — 使用宿主机网络（仅限 Linux）：\n> ```bash\n> docker run --network=host \\\n>   -v qwenpaw-data:\u002Fapp\u002Fworking \\\n>   -v qwenpaw-secrets:\u002Fapp\u002Fworking.secret \\\n>   agentscope\u002Fqwenpaw:latest\n> ```\n> 此时无需进行端口映射（`-p`），容器会直接共享宿主机网络。请注意，容器的所有端口都会暴露在宿主机上，如果该端口已被占用，可能会导致冲突。\n>\n> **注意**：如果您只挂载了 `\u002Fapp\u002Fworking` 而未单独创建 `\u002Fapp\u002Fworking.secret` 卷，则入口脚本会自动将密钥重定向到 `\u002Fapp\u002Fworking\u002F.secret`，从而确保密钥保存在同一卷中。\n\n该镜像是从零开始构建的。如需自行构建镜像，请参阅 `scripts\u002FREADME.md` 中的 [构建 Docker 镜像](scripts\u002FREADME.md#build-docker-image) 部分，然后推送到您的镜像仓库。\n\n---\n\n### 选项 4：部署在阿里云 ECS 上\n\n要在阿里云（ECS）上运行 QwenPaw，可以使用一键部署功能：打开 [QwenPaw 在阿里云（ECS）上的部署链接](https:\u002F\u002Fcomputenest.console.aliyun.com\u002Fservice\u002Finstance\u002Fcreate\u002Fcn-hangzhou?type=user&ServiceId=service-1ed84201799f40879884)，并按照提示操作。有关详细步骤，请参阅 [阿里云开发者：3 分钟部署你的 AI 助手](https:\u002F\u002Fdeveloper.aliyun.com\u002Farticle\u002F1713682)。\n\n---\n\n### 选项 5：使用 ModelScope\n\n**没有本地安装环境？** 可以通过 [ModelScope Studio](https:\u002F\u002Fmodelscope.cn\u002Fstudios\u002Ffork?target=AgentScope\u002FQwenPaw) 进行一键云端部署。请将 Studio 设置为 **非公开**，以防止他人控制您的 QwenPaw。\n\n---\n\n### 选项 6：桌面应用（Beta 版）\n\n> **Beta 版说明**：桌面应用目前处于 Beta 测试阶段，存在以下已知限制：\n> - **兼容性测试不完整**：尚未在所有系统版本和硬件配置上进行全面测试。\n> - **潜在性能问题**：启动时间、内存占用等性能方面可能仍需进一步优化。\n> - **功能开发中**：部分功能可能不稳定或缺失。\n\n如果您不熟悉命令行工具，可以直接下载并使用 QwenPaw 的桌面应用，而无需手动配置 Python 环境或执行命令。\n\n#### 下载\n\n您可以从 [GitHub Releases](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Freleases) 下载桌面应用：\n- **Windows**：`QwenPaw-Setup-\u003Cversion>.exe`\n- **macOS**：`QwenPaw-\u003Cversion>-macOS.zip`（推荐使用 Apple Silicon 设备）。\n\n#### 功能特点\n- ✅ **零配置**：下载后双击即可运行，无需安装 Python 或配置环境变量。\n- ✅ **跨平台**：支持 Windows 10 及以上版本和 macOS 14 及以上版本。\n- ✅ **可视化界面**：自动打开浏览器界面，无需手动输入地址。\n- ⚠️ **Beta 阶段**：功能仍在持续改进中，欢迎反馈。\n\n#### 首次启动\n**重要提示**：首次启动可能需要 10–60 秒（具体时间取决于您的系统配置）。应用程序需要初始化 Python 环境并加载依赖项，请耐心等待浏览器窗口自动打开。\n\n#### macOS：绕过系统安全限制\n当您从 Releases 下载 QwenPaw macOS 应用程序时，macOS 可能会显示：“Apple 无法验证‘QwenPaw’是否包含恶意软件”。这是因为该应用未经公证。您仍然可以通过以下方式打开：\n\n- **右键打开（推荐）**\n  右键单击（或按住 Control 键并单击）QwenPaw 应用程序 → **打开** → 在弹出的对话框中再次点击 **打开**。这会告知 Gatekeeper 您信任该应用；此后您可以像往常一样双击启动。\n\n- **在系统设置中允许**\n  如果仍然被阻止，请前往 **系统设置 → 隐私与安全性**，找到类似“QwenPaw 因来自未识别开发者而被阻止”的提示，并点击 **仍要打开** 或 **允许**。\n\n- **移除隔离属性（不推荐大多数用户使用）**\n  在终端中运行：\n  `xattr -cr \u002FApplications\u002FQwenPaw.app`\n  （或使用解压后的 `.app` 文件路径）。此操作会清除“从互联网下载”的隔离标记，通常可避免警告出现，但安全性及可控性不如使用“右键 → 打开”方法。\n\n有关详细的使用说明、故障排除及常见问题，请参阅 [桌面应用指南](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fdesktop)。\n\n---\n\n## API 密钥\n\n如果您使用 **云端大模型 API**（例如通义千问、Gemini、OpenAI），在开始聊天前必须配置 API 密钥。在设置有效的密钥之前，QwenPaw 将无法正常工作。详情请参阅[官方文档](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels)。\n\n**如何配置：**\n\n1. **控制台（推荐）** — 运行 `qwenpaw app` 后，打开 **http:\u002F\u002F127.0.0.1:8088\u002F** → **设置** → **模型**。选择提供商，输入 **API 密钥**，并启用该提供商和模型。\n2. **`qwenpaw init`** — 当您运行 `qwenpaw init` 时，它会引导您完成大模型提供商和 API 密钥的配置。按照提示选择提供商并输入您的密钥。\n3. **环境变量** — 对于 DashScope，您可以在 shell 中或工作目录下的 `.env` 文件中设置 `DASHSCOPE_API_KEY`。\n\n需要额外密钥的工具（例如用于网络搜索的 `TAVILY_API_KEY`）可以在控制台的 **设置 → 环境变量** 中进行设置，详情请参阅[配置](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconfig)。\n\n> **仅使用本地模型？** 如果您使用[本地模型](#local-models)（llama.cpp \u002F Ollama \u002F LM Studio），则**无需**任何 API 密钥。\n\n## 本地模型\n\nQwenPaw 可以在您的设备上完全运行大模型，无需任何 API 密钥或云服务。详情请参阅[官方文档](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels)。\n\n| 后端       | 最适合                                 | 安装                                                              |\n| ------------- | ---------------------------------------- | -------------------------------------------------------------------- |\n| **llama.cpp** | 跨平台（macOS \u002F Linux \u002F Windows） | 无需额外安装；在网页界面上点击“下载 Llama.cpp”。 |\n| **Ollama**    | 跨平台（需安装 Ollama 服务） | 提前安装并启动 Ollama 应用程序。 |\n| **LM Studio** | 跨平台（需安装 LM Studio 服务） | 提前安装并启动 LM Studio 应用程序。 |\n\n---\n\n## 文档\n\n| 主题                                                                 | 描述                                      |\n| --------------------------------------------------------------------- | ------------------------------------------------ |\n| [简介](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fintro)                | QwenPaw 是什么以及如何使用它                  |\n| [快速入门](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fquickstart)            | 安装并运行（本地或 ModelScope Studio）    |\n| [控制台](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconsole)                   | 网页界面：聊天和智能体配置            |\n| [模型](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmodels)                     | 配置云端、本地和自定义提供商    |\n| [渠道](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels)                  | 钉钉、飞书、QQ、Discord、iMessage 等 |\n| [技能](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fskills)                      | 扩展和自定义功能               |\n| [MCP](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmcp)                            | 管理 MCP 客户端                               |\n| [记忆](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmemory)                     | 长期记忆                     |\n| [上下文](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcontext)                   | 上下文管理机制                     |\n| [魔法命令](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcommands)           | 在无需等待 AI 回答的情况下控制对话状态 |\n| [心跳](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fheartbeat)                | 定期签到和摘要                    |\n| [多智能体](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fmulti-agent)           | 创建多个智能体并实现协作  |\n| [配置与工作目录](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fconfig) | 工作目录和配置文件                |\n| [CLI](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fcli)                            | 初始化、定时任务、技能、清理                   |\n| [常见问题解答](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Ffaq)                           | 常见问题及故障排除             |\n\n完整文档位于此仓库：[website\u002Fpublic\u002Fdocs\u002F](website\u002Fpublic\u002Fdocs\u002F)。\n\n---\n\n## 安全特性\n\nQwenPaw 包含多层安全机制，以保护您的数据和系统：\n\n- **工具防护** — 自动拦截危险的 Shell 命令（如 `rm -rf \u002F`、fork 恶意软件、反向 Shell 等）。\n- **文件访问防护** — 限制智能体对敏感路径的访问（如 `~\u002F.ssh`、密钥文件、系统目录等）。\n- **技能安全扫描** — 在安装技能前自动扫描，检测诸如提示注入、命令注入、硬编码密钥、数据外泄等风险。\n- **本地部署** — 所有数据和内存均存储在本地，不会上传至第三方（当使用云端大模型 API 时，对话内容会发送到相应的 API 提供商）。\n- **Web 认证** — 控制台可选登录保护。默认关闭；设置 `QWENPAW_AUTH_ENABLED=true` 即可启用。详情请参阅[Web 认证](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fsecurity#Web-Authentication)。\n\n详细信息请参阅[安全文档](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fsecurity)。\n\n---\n\n## 常见问题解答\n\n有关常见问题、故障排除技巧及已知问题，请访问**[FAQ 页面](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Ffaq)**。\n\n---\n\n## 保持关注\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_19d1165bd63e.gif\" width=\"600\" alt=\"Star QwenPaw\" \u002F>\n\u003C\u002Fa>\n\n在 GitHub 上为 QwenPaw 点赞，即可第一时间获取新版本通知。\n\n---\n\n## Roadmap\n\n| Area                                  | Item                                                                                                                                             | Status               |\n| ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------- |\n| **Horizontal Expansion**              | More channels, models, skills, MCPs — **community contributions welcome**                                                                        | Seeking Contributors |\n| **Existing Feature Extension**        | Display optimization, download hints, Windows path compatibility, etc. — **community contributions welcome**                                     | Seeking Contributors |\n| **Console Web UI**                    | Expose more info\u002Fconfig in the Console                                                                                                           | In Progress          |\n| **Multi-agent**                      | Agentic Ralph Loop                                                                                  | In Progress          |\n| **Multimodal**                        | Voice\u002Fvideo calls and real-time interaction                                                                                                      | In Progress          |\n| **Small + Large Model Collaboration** | Multi-model routing; different models for different tasks                                                                                          | In Progress          |\n| **Memory System**                     | Experience distillation & skill extraction                                                                                                       | In Progress          |\n|                                       | Memory mechanism switching                                                                                                                       | In Progress             |\n|                                       | Multimodal memory fusion                                                                                                                         | Planned              |\n|                                       | Context-aware proactive delivery                                                                                                                 | Planned              |\n| **Sandbox**                           | Deeper integration with AgentScope Runtime sandboxes                                                                                             | In Progress          |\n| **Cloud-native**                      | Deeper integration with AgentScope Runtime; leverage cloud compute, storage, tools, and skills                                                         | In Progress              |\n| **Skills Hub**                        | Enrich the [AgentScope Skills](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope-skills) repository and improve discoverability of high-quality skills | Planned              |\n\n\n_Status:_ **In Progress** — actively being worked on; **Planned** — queued or under design, also welcome contributions; **Seeking Contributors** — we strongly encourage community contributions.\n\n## Install from source\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw.git\ncd QwenPaw\n\n# Build console frontend first (required for web UI)\ncd console && npm ci && npm run build\ncd ..\n\n# Copy console build output to package directory\nmkdir -p src\u002Fqwenpaw\u002Fconsole\ncp -R console\u002Fdist\u002F. src\u002Fqwenpaw\u002Fconsole\u002F\n\n# Install Python package\npip install -e .\n```\n\n- **Dev** (tests, formatting): `pip install -e \".[dev,full]\"`\n- **Then**: Run `qwenpaw init --defaults`, then `qwenpaw app`.\n\n> **Note for updates:** When updating to a new major version after `git pull`, please also rebuild the frontend, reinstall the package (`pip install -e .`), restart `qwenpaw app`, and clear your browser cache with `Ctrl+Shift+R` (or `Cmd+Shift+R` on macOS).\n\n---\n\n## Contributing\n\nQwenPaw evolves through open collaboration, and we welcome all forms of contribution! Check the [Roadmap](#roadmap) above (especially items marked **Seeking Contributors**) to find areas that interest you, and read [CONTRIBUTING](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fblob\u002Fmain\u002FCONTRIBUTING.md) to get started. We particularly welcome:\n\n- **Horizontal expansion** — new channels, model providers, skills, MCPs.\n- **Existing feature extension & refinement** — display and interaction improvements, download hints, Windows path compatibility, etc.\n\nJoin [GitHub Discussions](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fdiscussions) to discuss ideas or pick up tasks.\n\n---\n\n## Why QwenPaw?\n\nQwenPaw stands for Qwen Personal Agent Workstation, and also embodies the wisdom of Qwen and the warmth of a Paw.\nWe hope it is not a cold tool, but an intelligent and warm \"little paw\" always ready to help—a most intuitive partner in your digital life.\n\n---\n\n## Built by\n\n[AgentScope team](https:\u002F\u002Fgithub.com\u002Fagentscope-ai) · [AgentScope](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope) · [AgentScope Runtime](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002Fagentscope-runtime) · [ReMe](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FReMe)\n\n---\n\n## Contact us\n\n| [Discord](https:\u002F\u002Fdiscord.gg\u002FeYMpfnkG8h)                     | [X (Twitter)](https:\u002F\u002Fx.com\u002Fagentscope_ai)                   | [DingTalk](https:\u002F\u002Fqr.dingtalk.com\u002Faction\u002Fjoingroup?code=v1,k1,OmDlBXpjW+I2vWjKDsjvI9dhcXjGZi3bQiojOq3dlDw=&_dt_no_comment=1&origin=11) | [RedNote](https:\u002F\u002Fwww.xiaohongshu.com\u002Fuser\u002Fprofile\u002F691c18db0000000037032be9) |\n| ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |\n| [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_5e73a7413fa9.png\" width=\"80\" height=\"80\" alt=\"Discord\">](https:\u002F\u002Fdiscord.gg\u002FeYMpfnkG8h) | [\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-black.svg?logo=x&logoColor=white\" width=\"80\" height=\"80\" alt=\"X\">](https:\u002F\u002Fx.com\u002Fagentscope_ai) | [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_584abd3b9a38.png\" width=\"80\" height=\"80\" alt=\"DingTalk\">](https:\u002F\u002Fqr.dingtalk.com\u002Faction\u002Fjoingroup?code=v1,k1,OmDlBXpjW+I2vWjKDsjvI9dhcXjGZi3bQiojOq3dlDw=&_dt_no_comment=1&origin=11) | [\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_1e0690be8700.jpg\" width=\"80\" height=\"80\" alt=\"RedNote\">](https:\u002F\u002Fwww.xiaohongshu.com\u002Fuser\u002Fprofile\u002F691c18db0000000037032be9) |\n\n---\n\n## Telemetry\n\nQwenPaw collects **anonymous** usage data during `qwenpaw init` to help us understand our user base and prioritize improvements. Data is sent **once per version** — when you upgrade QwenPaw, telemetry is re-collected so we can track version adoption.\n\n**What we collect:**\n\n- QwenPaw version (e.g., 0.0.7)\n- Install method (pip, Docker, or desktop app)\n- OS and version (e.g., macOS 14.0, Ubuntu 22.04)\n- Python version (e.g., 3.13)\n- CPU architecture (e.g., x86_64, arm64)\n- GPU availability (yes\u002Fno)\n\n**What we do NOT collect:** No personal data, no files, no credentials, no IP addresses, no identifiable information.\n\nWhen running `qwenpaw init` interactively, you will be asked whether to opt in. If you choose `--defaults`, telemetry is accepted automatically. The prompt appears once per version and never affects QwenPaw's functionality.\n\n---\n\n## License\n\nQwenPaw is released under the [Apache License 2.0](LICENSE).\n\n---\n\n## Contributors\n\nAll thanks to our contributors:\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_CoPaw_readme_2751d8b17041.png\" alt=\"Contributors\" \u002F>\n\u003C\u002Fa>","# QwenPaw 快速上手指南\n\nQwenPaw 是一款可本地部署、支持多通道连接的个人 AI 助手。数据完全由你掌控，支持技能扩展与多智能体协作。\n\n## 1. 环境准备\n\n### 系统要求\n- **操作系统**：Windows (10\u002F11), macOS, Linux\n- **Python 版本**：3.10 ~ 3.13 (推荐 3.11+)\n- **网络环境**：需能访问 GitHub 和 PyPI（若受限请使用脚本安装或国内镜像）\n\n### 前置依赖\n- **方案 A（推荐）**：无需手动安装 Python，使用官方一键脚本自动管理环境。\n- **方案 B**：若自行管理 Python，请确保已安装 `pip` 和 `git`。\n\n> **注意**：Windows 企业版用户若遇到 PowerShell 执行限制，建议使用 CMD 运行 `.bat` 脚本或手动配置环境变量（详见原文注意事项）。\n\n---\n\n## 2. 安装步骤\n\n你可以选择以下任意一种方式进行安装：\n\n### 方式一：一键脚本安装（推荐新手）\n无需预先安装 Python，脚本会自动下载包管理器 `uv`、创建虚拟环境并安装所有依赖。\n\n**macOS \u002F Linux:**\n```bash\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.sh | bash\n```\n\n**Windows (CMD):**\n```CMD\ncurl -fsSL https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.bat -o install.bat && install.bat\n```\n\n**Windows (PowerShell):**\n```powershell\nirm https:\u002F\u002Fqwenpaw.agentscope.io\u002Finstall.ps1 | iex\n```\n\n> **提示**：若需支持本地模型（如 Ollama），可在命令后添加 `--extras ollama` 参数。\n\n### 方式二：pip 安装（适合开发者）\n如果你已经配置好 Python 环境：\n\n```bash\npip install qwenpaw\n```\n\n---\n\n## 3. 基本使用\n\n安装完成后，只需三条命令即可启动服务：\n\n### 第一步：初始化配置\n生成默认配置文件和数据目录。\n```bash\nqwenpaw init --defaults\n```\n*注：若需交互式配置，可运行 `qwenpaw init`。*\n\n### 第二步：启动应用\n```bash\nqwenpaw app\n```\n\n### 第三步：开始对话\n启动成功后，终端会显示访问地址。\n1. 打开浏览器访问：**http:\u002F\u002F127.0.0.1:8088\u002F**\n2. 在控制台页面配置你的 **API Key**（或使用本地模型）。\n3. 直接在网页控制台进行对话测试。\n\n---\n\n## 4. 进阶指引\n\n- **接入钉钉\u002F飞书\u002F微信**：完成上述步骤后，参考 [通道配置文档](https:\u002F\u002Fqwenpaw.agentscope.io\u002Fdocs\u002Fchannels) 绑定机器人。\n- **使用本地模型**：无需 API Key，下载模型后在配置中选择本地推理后端（如 Ollama）。\n- **卸载工具**：\n  - 保留配置卸载：`qwenpaw uninstall`\n  - 彻底清除（含数据）：`qwenpaw uninstall --purge`","自由职业数据分析师李明需要同时处理多个客户的日报生成、文档摘要及日程提醒任务，且对数据隐私有极高要求。\n\n### 没有 QwenPaw 时\n- **数据泄露风险高**：不得不将敏感客户数据上传至第三方公有云 AI 平台，时刻担忧隐私合规问题。\n- **工具割裂效率低**：需要在网页版聊天窗口、本地脚本和日历应用间反复切换，无法在一个界面完成“读取 PDF-分析数据-发送报告”的闭环。\n- **功能扩展门槛高**：想要增加“自动抓取行业新闻”或“定制 Excel 处理”功能时，必须重写大量底层代码，难以快速响应客户需求。\n- **记忆上下文丢失**：不同客户的对话历史和偏好设置分散在各个浏览器标签页中，每次重启都需要重新交代背景信息。\n\n### 使用 QwenPaw 后\n- **数据完全自主可控**：李明将 QwenPaw 部署在本地服务器，所有客户数据仅在本地流转，彻底消除了上传第三方带来的隐私顾虑。\n- **多通道一站式协作**：通过 QwenPaw 统一接入多个聊天应用，直接调用内置的 PDF 处理和新闻摘要技能，在一个对话框内即可完成从文件解析到报告生成的全流程。\n- **技能按需热插拔**：利用其易扩展架构，李明轻松加载了自定义的“财经数据清洗”插件，无需修改核心代码即可让助手掌握新技能。\n- **个性化记忆持久化**：QwenPaw 为每个客户维护独立的记忆单元，自动关联历史对话与偏好，再次交互时无需重复指令，服务更加连贯智能。\n\nQwenPaw 让李明在确保数据绝对安全的前提下，拥有了一个可无限生长、懂业务且高度集成的私人智能工作流中枢。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fagentscope-ai_QwenPaw_db5d3c5f.png","agentscope-ai","AgentScope-AI","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fagentscope-ai_aa2b4933.png","",null,"agentscope.team@gmail.com","https:\u002F\u002Fgithub.com\u002Fagentscope-ai",[80,84,88,92,96,100,104,107,111,114],{"name":81,"color":82,"percentage":83},"Python","#3572A5",70.7,{"name":85,"color":86,"percentage":87},"TypeScript","#3178c6",22.4,{"name":89,"color":90,"percentage":91},"Less","#1d365d",3.9,{"name":93,"color":94,"percentage":95},"CSS","#663399",1.2,{"name":97,"color":98,"percentage":99},"PowerShell","#012456",0.7,{"name":101,"color":102,"percentage":103},"Shell","#89e051",0.5,{"name":105,"color":106,"percentage":103},"Batchfile","#C1F12E",{"name":108,"color":109,"percentage":110},"JavaScript","#f1e05a",0.1,{"name":112,"color":113,"percentage":110},"Dockerfile","#384d54",{"name":115,"color":76,"percentage":116},"NSIS",0,15227,2055,"2026-04-14T00:00:40","Apache-2.0","Linux, macOS, Windows","未说明（支持本地模型部署，具体需求取决于所选模型；若使用 Ollama 等本地推理引擎，需参考对应模型的硬件要求）","未说明",{"notes":125,"python":126,"dependencies":127},"提供多种安装方式：pip 安装、一键脚本安装（自动配置 uv 虚拟环境及 Node.js）、桌面应用测试版。支持通过 Ollama 运行本地模型无需 API Key。Windows 企业版 LTSC 用户需注意 PowerShell 受限语言模式可能导致环境变量配置失败，需手动添加 uv 和 QwenPaw 路径到系统 Path。数据可完全本地部署以保护隐私。","3.10 ~ \u003C3.14",[128,129],"uv (Python 包管理器)","Node.js (前端资源依赖)",[13,35],[132,133,134,135,136,137],"agent","llm-tools","llms","skills","agentscope","agentscope-runtime","2026-03-27T02:49:30.150509","2026-04-14T12:27:57.134626",[141,146,151,156,161,166],{"id":142,"question_zh":143,"answer_zh":144,"source_url":145},33159,"如何从 CoPaw 平滑升级到 QwenPaw？需要迁移数据吗？","不需要做额外的数据迁移工作。直接安装 QwenPaw 即可自动升级。系统会优先使用之前的默认工作目录（~\u002F.copaw），保证原有数据（如自建智能体、技能、配置、记忆等）可复用。原有的 CLI 命令仍然有效，可以使用 `copaw` 或 `qwenpaw` 启动和运行。如果是脚本安装，停止旧服务后执行新脚本安装，重启电脑后启动即可。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F3288",{"id":147,"question_zh":148,"answer_zh":149,"source_url":150},33160,"更新版本后出现 'Authentication Fails' 报错怎么办？","这通常是因为模型配置在更新后失效或未正确加载。请尝试重新配置模型：在界面中即使默认显示了模型（如 deepseek-chat），也请手动重新选择一遍该模型并保存配置，以刷新认证信息。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F2102",{"id":152,"question_zh":153,"answer_zh":154,"source_url":155},33161,"CoPaw 进程在空闲时 CPU 占用率持续 100% 如何解决？","这是一个已知的忙轮询（busy-polling）问题。临时解决方法是修改源码中的监听时间间隔（例如从 2 秒改为 30 秒）并重启服务。官方已在后续版本中修复了相关异步事件循环的问题，建议升级到最新版本以获得永久修复。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F2218",{"id":157,"question_zh":158,"answer_zh":159,"source_url":160},33162,"安装新版本后启动报错 'ImportError: cannot import name ServerError' 怎么办？","这通常是由于依赖冲突或环境缓存导致的。最有效的解决方法是创建一个新的虚拟环境并重新安装。如果使用 Docker，请注意 `latest` 标签仅对应稳定版，若要测试最新的 Beta 版本（如 v0.1.0b2），需使用 `pre` 标签或直接通过 pip 指定版本安装：`pip install copaw==0.1.0b2`。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F1682",{"id":162,"question_zh":163,"answer_zh":164,"source_url":165},33163,"为什么使用 'pip install --upgrade copaw' 更新后，界面版本号没有变化？","`copaw update` 命令仅在 v0.1.0b1 及更高版本中支持。对于旧版本（如 0.0.x），pip 更新可能无法正确刷新前端资源或命令不可用。建议先执行 `copaw uninstall` 卸载旧版本，然后重新安装最新版本，或者直接下载包含最新前端的完整安装包。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F1195",{"id":167,"question_zh":168,"answer_zh":169,"source_url":170},33164,"更新版本后自定义模型 API 配置消失或无法保存怎么办？","这是旧版本（如 v0.0.4）的一个已知 Bug，导致配置文件读取或写入异常。该问题已在后续的 PR（#345 和 #346）中修复。请等待并发布最新的 Beta 版本或稳定版进行升级，升级后自定义模型配置将恢复正常，无需手动编辑配置文件。","https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fissues\u002F398",[172,177,182,187,192,197,202,207,212,217,222,227,232,237,242,247,252,257,262,267],{"id":173,"version":174,"summary_zh":175,"released_at":176},255331,"v1.1.1-beta.1","## 变更内容\n* docs(news): 由 @xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3287 中更新新闻\n* chore: 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3289 中更新 IP 标志\n* chore(version): 由 @xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3304 中将版本号提升至 1.1.1b1\n* fix(console): 由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3294 中修复模型设置页面，并将默认 LLM 标题移至卡片中\n* CI: 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3307 中自动添加“待评审”标签\n* fix: 由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3186 中实现 URL 新标签打开\n* fix: 由 @ltzu929 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3267 中修复聊天会话置顶按钮在暗黑模式下对比度过低的问题\n* feat(skill): 由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3310 中添加对 Clawhub 格式请求的支持\n* perf(QQ): 由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3188 中通过强制关闭 WebSocket 的方式修复 QQ 通道 stop() 阻塞 8 秒的问题\n* feat(Feishu): 由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3207 中为 Feishu 通道添加引用消息处理功能\n* feat(console): 由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3175 中为支持模型发现的模型提供商添加模型 ID 选择的自动补全下拉菜单\n* ci: 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3316 中修复“待评审”的 CI 流程\n* refactor(DingTalk): 由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3236 中将 API 调用迁移至官方 SDK，并添加表情反应反馈\n* Console\u002Fchannel: 由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3315 中进行相关调整\n* Fix(Local Provider): 由 @pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3321 中修复 Windows 桌面端出现的“下载失败：[Errno 22] 无效参数”问题\n* chore (ci): 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3324 中修复带有“Review Opt”选项的“待评审”CI 流程\n* feat(matrix): 由 @maplefeng-a 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F2509 中为 Matrix 通道增强 E2EE、提及功能及历史记录支持（#2433）\n* fix(skill): 由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3320 中修复技能创建时的加载问题以及选择行为\n* chore: 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3326 中修复“待评审”的 CI 流程\n* feat(provider): 由 @seoeaa 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F1192 中添加 OpenRouter 提供商，支持 HTTP-Referer 和 X-Title 头\n* feat(skill): 由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3330 中更新技能命令描述，并补充文档\n* fix(website UI): 由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3332 中优化贡献者样式\n* feat(security): 由 @gnipping 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F3303 中添加 CC 规则\n* feat(provider): 由 @lhpqaq 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FQwenPaw\u002Fpull\u002F2463 中支持与 OpenCode 的集成\n* feat","2026-04-13T13:09:49",{"id":178,"version":179,"summary_zh":180,"released_at":181},255332,"v1.1.0","## CoPaw 正式更名为 QwenPaw\n\n此次品牌升级标志着我们迈入开源发展的新阶段，迈出了重要一步。\n\n新名称更贴切地体现了我们正在构建的开放生态，以及我们持续推进的更广阔的发展方向：\n\n_Qwen_ — 代表与通义开源生态的深度融合，更加聚焦模型层，包括面向智能体的本地化模型，以及大小模型之间的智能协同。\n\n_Paw_ — 沿承我们最初的使命——陪伴用户，成为值得用户信赖的个人助手。\n\n此次品牌升级不会改变我们在开源社区中的愿景与目标：\n\n- 打造更实用、更安全、更个性化的AI助手；\n- 坚持开源协作，与社区携手共进。\n\n感谢大家一路以来的支持。从今天起，QwenPaw 将继续与您同行，共同成长！\n\n更多更新即将发布。","2026-04-12T08:57:12",{"id":183,"version":184,"summary_zh":185,"released_at":186},255333,"v1.1.0-beta.1","## 变更内容\n* 重构（环境）：从 COPAW 迁移到 QWENPAW 环境变量… 由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3171 中完成\n* 杂项：重命名脚本及 CI 模块 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3169 中完成\n* 杂项：在 CI 中添加手动 PyPI 发布触发器 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3172 中完成\n* 重构（核心）：将项目名称由 copaw 更改为 qwenpaw 由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3180 中完成\n* 重构（CLI）：将所有出现的“CoPaw”字样更改为“QwenPaw” 由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3189 中完成\n* 杂项（频道）：在频道模块中进行重命名 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3193 中完成\n* 重构（安全）：重命名 copaw 相关引用，并保留兼容性回退 由 @gnipping 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3192 中完成\n* 重构（路由）：将 copaw 标识符更改为 qwenpaw 由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3194 中完成\n* 重构（代理）：进行重命名 由 @lalaliat 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3196 中完成\n* 重构（工具）：在工具模块中将 CoPaw 更名为 QwenPaw 由 @jinglinpeng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3203 中完成\n* 重构（内存）：重命名 token 计数函数，并更新内存管理 由 @jinliyl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3199 中完成\n* 杂项（插件）：重命名插件 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3205 中完成\n* 重构（技能）：为技能模块进行重命名 由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3201 中完成\n* 重构：为应用运行器和工具库进行重命名 由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3202 中完成\n* 重构（文档）：将文档中的名称更改为 QwenPaw 由 @cuiyuebing 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3200 中完成\n* 控制台\u002F性能优化 由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3210 中完成\n* 重构（网站）：将网站名称由 copaw 更改为 qwenpaw 由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3212 中完成\n* 重构：在注释中进行重命名 由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3213 中完成\n* 重构（配置）：在配置中将 copaw 更改为 qwenpaw 由 @yuanxs21 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3216 中完成\n* 重构（工具库）：将 copaw 工具库更名为 qwenpaw 工具库 由 @x1n95c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3204 中完成\n* 杂项：在工作区模块中进行小幅修复 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3217 中完成\n* 重构（根目录）：在根目录中将 copaw 更改为 qwenpaw 由 @yuanxs21 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3220 中完成\n* 重构（安全）：使用旧版变量 由 @gnipping 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3223 中完成\n* 重构（本地模型、提供商）：将 copaw 更改为 qwenpaw 由 @garyzhang99 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3195 中完成\n* 重构（网站）：进行重命名 由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3219 中完成\n* 杂项：异步安全读取文件 由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3228 中完成\n* 更新","2026-04-12T08:05:34",{"id":188,"version":189,"summary_zh":190,"released_at":191},255334,"v1.0.2","## ✨ 新增\r\n\r\n**核心功能**\r\n\r\n- **插件**：从工作区的 `plugins\u002F` 文件夹安装扩展程序（[#3101](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3101)、[#3131](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3131)、[#3132](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3132)）。\r\n- **`copaw task`**：无需 Web 服务器，直接在终端运行一次性任务（[#3031](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3031)）。\r\n- **聊天中的 `\u002Fmodel`**：无需打开设置即可切换模型、查看可用模型列表、重置为默认模型以及检查模型信息（[#3133](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3133)）。\r\n\r\n**模型与提供商**\r\n\r\n- **SiliconFlow**：提供 SiliconFlow 兼容 OpenAI API 的即用型中国及国际端点（[#2886](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2886)）。\r\n- **CoPaw Local**：支持图像和视频的模型、更丰富的控制台设置，以及在 Windows 上更可靠地完成下载和能力检测（[#3021](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3021)、[#3087](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3087)、[#3140](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3140)）。\r\n\r\n**安全**\r\n\r\n- **更安全的保存密钥**：API 密钥等敏感值在磁盘上加密存储；在可能的情况下，加密密钥会保留在操作系统密钥链中（[#3025](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3025)）。\r\n\r\n**控制台与 UI**\r\n\r\n- **聊天输入历史**：使用方向键浏览之前的用户输入内容（[#2466](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2466)）。\r\n- **聊天搜索**：跨会话搜索消息文本（[#2842](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2842)）。\r\n- **置顶会话**：将线程置顶，使其始终位于列表顶部（[#3137](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3137)）。\r\n- **按代理聊天**：切换代理时恢复上次打开的聊天窗口（[#3155](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3155)）。\r\n- **视觉优化**：内置工具图标以及模型旁边的提供商标志（[#3061](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3061)、[#3130](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3130)）。\r\n\r\n**技能与工具**\r\n\r\n- **技能命令**：`\u002Fskills` 可显示当前频道已启用的技能；输入 `\u002F技能名` 即可打开或随消息一起执行该技能（[#3150](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3150)）。\r\n- **池技能标签**：通过标签对共享技能池进行分类管理（[#2837](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2837)、[#3069](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3069)）。\r\n- **MCP 工具发现**：通过 HTTP API 查询每个连接的 MCP 服务器，获取其工具名称、描述和参数信息（[#3149](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3149)）。\r\n\r\n**渠道**\r\n\r\n- **QQ**：在不同消息类型中一致地发送文件和富媒体内容（[#3012](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3012)）。\r\n- **企业微信**：传入消息会保留引用\u002F回复上下文，以便代理能够看到所引用的内容（[#3024](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3024)）。\r\n\r\n## 🔄 ","2026-04-09T12:52:06",{"id":193,"version":194,"summary_zh":195,"released_at":196},255335,"v1.0.2-beta.2","## 变更内容\n* 修复：@xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3112 中对网站进行的热修复\n* 撤销：“修复：热修复网站”——@xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3115 中执行\n* 撤销：“网站性能优化”——@xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3116 中执行\n* 重构（控制台）：使用 pretty-bytes 统一文件大小格式——@bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2808 中完成\n* 性能优化（控制台）：使用 React.lazy 对非默认路由页面进行懒加载——@yyyzl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3122 中实现\n* 文档：修复插件中的拼写错误——@qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3131 中完成\n* 功能（控制台）：在模型选择器上添加模型提供商图标——@bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3130 中实现\n* 功能（日志）：添加 PlainFormatter 以改进日志格式——@zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3147 中完成\n* 功能：性能优化——@zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3141 中实现\n* 功能：新增 `\u002Fmodel` 命令——@rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3133 中实现\n* 杂项（版本）：将版本号提升至 1.0.2b2——@xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3156 中完成\n* 功能（聊天）：为聊天会话添加置顶功能——@zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3137 中实现\n* 文档：修复插件 MD 文件中的代码——@qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3132 中完成\n* 功能（会话）：实现聊天会话展示——@qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3155 中完成\n* 网站构建优化——@yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3138 中完成\n* 修复（Provider）：修复 Windows 桌面版下载及 Copaw 本地多模态探测问题——@pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3140 中完成\n* 修复（飞书）：为每个代理隔离事件循环，防止跨实例干扰——@hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3145 中完成\n* 性能优化：将悬停状态移至卡片组件，并添加 React.memo——@zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3158 中完成\n* 修复：在 shell 工具 _collapse_embedded_newlines 中保留带引号字符串内的换行符——@hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3070 中完成\n* 功能（技能）\u002F命令：新增 \u002Fskills 列表显示及 \u002F\u003C技能> 调用命令——@Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3150 中实现\n* 功能（MCP）：支持为 MCP 列出工具——@qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3149 中完成\n* 功能（错误码）：更新错误码——@xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3110 中完成\n* 功能：修复 HTTP MCP——@rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3167 中完成\n\n## 新贡献者\n* @yyyzl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3122 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.2-beta.1...v1.0.2-beta.2","2026-04-09T12:18:08",{"id":198,"version":199,"summary_zh":200,"released_at":201},255336,"v1.0.2-beta.1","## 变更内容\n* 新增（控制台）：仅限本地模型提供商进行模型发现，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2938 中实现\n* 构建（版本）：将版本号升级至 1.0.2b1，由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2942 中完成\n* 新增（QQ）：为所有消息类型完善富媒体文件发送功能，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3012 中实现\n* 新增（提供商）：CoPaw Local 支持多模态功能，由 @pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3021 中实现\n* 新增（技能池分类标签）：由 @mautops 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2837 中完成\n* 修复（控制台）：本地化测试连接消息，由 @hikariming 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2913 中修复\n* 新增（提供商）：新增 SiliconFlow CN 和 SiliconFlow 国际版的模型提供商，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2886 中实现\n* 修复（浏览器）：修复重复元素引用问题，并完善浏览器扩展文档，由 @x1n95c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3023 中完成\n* 修复（Discord）：在 _chunk_text 中为代码块结束符预留空间，由 @energyd 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2976 中修复\n* 新增（Minimax）：将 Minimax 国际版的 support_connection_check 设置为 False，由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3032 中实现\n* 新增（企业微信）：支持解析传入消息中的引用消息，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3024 中实现\n* 新增：在所有聊天会话中添加搜索功能，由 @fengrui-z 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2842 中实现\n* 修复（iMessage）：为 iMessage 渠道添加缺失的 dm_policy 和白名单检查，由 @alostbear 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2491 中修复\n* 新增（移除技能分类）：由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3069 中实现\n* 重构（Web）：时区选择选项支持国际化显示，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2497 中完成\n* 修复（内存）：使 force_memory_search 的超时时间可配置，由 @octo-patch 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3057 中修复\n* 新增（安全）：使用 Fernet 和操作系统密钥链对静止状态下的秘密进行加密，由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3025 中实现\n* 新增（控制台）：为内置工具添加图标，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3061 中实现\n* 修复（技能管理器）：确保元数据为字典后再进行访问…，由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3072 中修复\n* 网站性能优化，由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3067 中完成\n* 新增（聊天）：在聊天页面上添加终端风格的消息历史导航功能，可通过方向键操作，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2466 中实现\n* 控制台\u002F技能，由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3081 中实现\n* 修复（钉钉）：在回退方案中，通过 Open API 而非纯文本路径发送媒体文件，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F3060 中修复\n* 新增：支持 u","2026-04-08T13:52:06",{"id":203,"version":204,"summary_zh":205,"released_at":206},255337,"v1.0.1","## ✨ 新增\r\n\r\n**模型与提供商**\r\n\r\n- **智谱模型提供商**：内置对智谱AI模型的支持（[#2858](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2858)）\r\n- **多模态模型的视频分析功能**：扩展了多模态支持，可自动提取和分析视频文件（[#2627](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2627)）\r\n- **模型特定的生成参数**：通过模型设置实现针对不同模型的生成关键字参数配置（[#2892](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2892)）\r\n- **CoPaw Local 更新支持**：为CoPaw Local 提供商引入自动更新机制，包含版本检查和下载功能（[#2889](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2889)）\r\n- **宽松的工具调用解析器**：放宽对工具调用的解析规则，以处理格式不规范的JSON，并提升与各类大模型输出的兼容性（[#2832](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2832)）\r\n\r\n**控制台与UI**\r\n\r\n- **智能体拖拽排序功能**：通过拖拽界面实现智能体顺序的持久化保存（[#2695](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2695)）\r\n- **聊天会话状态指示器**：为每个聊天会话添加可视化状态指示，显示其当前是否处于活动状态（[#2803](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2803)）\r\n- **首选会话优先级**：自动将首选聊天会话移至会话列表顶部，方便快速访问（[#2864](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2864)）\r\n- **系统深色模式选项**：深色模式切换新增系统偏好选项，可自动根据系统主题进行切换（[#2678](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2678)）\r\n- **默认智能体自动切换功能**：当所选智能体被删除或禁用时，智能体选择器会自动切换到默认智能体（[#2640](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2640)）\r\n\r\n**技能与工具**\r\n\r\n- **技能批量操作功能**：支持在多个智能体之间批量删除、下载及广播技能（[#2743](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2743)）\r\n- **智能体禁用时停止服务**：当智能体被禁用时，自动停止其后台服务以释放资源（[#2746](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2746)）\r\n- **技能依赖项列表格式支持**：在技能依赖项字段中，除了字符串格式外，新增了列表格式的支持（[#2504](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2504)）\r\n\r\n\r\n**渠道**\r\n\r\n- **OneBot v11 渠道**：新增NapCat\u002FQQ渠道集成，采用反向WebSocket服务器，全面支持QQ协议，包括个人账号和群消息功能（[#2870](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2870)）\r\n- **钉钉AI卡片在工作区跟踪器中**：在工作区跟踪器路径中新增AI卡片支持，并重构了共享核心模块（[#2741](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2741)）\r\n- **飞书DONE反应功能**：新增流程完成钩子，支持在飞书工作区路径中使用DONE反应表情（[#2727](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2727)）\r\n- *","2026-04-03T18:47:27",{"id":208,"version":209,"summary_zh":210,"released_at":211},255338,"v1.0.1-beta.2","## 变更内容\n* 功能（控制台）：将首选聊天会话移至会话列表顶部，由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2864 中实现。\n* 修复（工具）：浏览器使用时的空闲看门狗自动取消，由 @x1n95c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2843 中修复。\n* 功能（模型）：添加智谱AI提供商，由 @hikariming 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2858 中实现。\n* 修复：通过 Google Fonts 加载 Inter、Lato 和 Newsreader 字体，由 @octo-patch 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2867 中完成。\n* 功能（企业微信）：用服务器端二维码替换 JS SDK 弹出窗口，并重构为通用提供商模式，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2891 中实现。\n* 功能（模型）：支持为每个模型修改生成参数，由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2892 中实现。\n* 功能（提供商）：支持 CoPaw 本地更新，由 @pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2889 中实现。\n* 修复（提供商）：修复命令运行器路径问题，由 @pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2905 中修复。\n* 修复（聊天）：保留已重命名的会话标题，由 @fancyboi999 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2847 中修复。\n* 修复：为 Windows 浏览器启动添加 --no-sandbox 和 --disable-gpu 参数，由 @rikey123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2861 中完成。\n* 功能（渠道）：为 NapCat\u002FQQ 集成添加 OneBot v11 渠道，由 @solarhell 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2870 中实现。\n\n## 新贡献者\n* @rikey123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2861 中完成了首次贡献。\n* @solarhell 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2870 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.1-beta.1...v1.0.1-beta.2","2026-04-03T14:03:53",{"id":213,"version":214,"summary_zh":215,"released_at":216},255339,"v1.0.1-beta.1","## 变更内容\n* chore(version)：由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2795 中将版本更新为 1.0.1b1\n* fix(console)：由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2793 中修复工作区中文件时间戳显示为“NaNd ago”的问题\n* Conole\u002Fskill：由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2794 中完成\n* feat(chat)：由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2803 中为聊天会话添加状态指示器，并增强本地化支持\n* remove none used package-lock.json：由 @mautops 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2791 中移除未使用的 package-lock.json 文件\n* feat(memory)：由 @YingchaoX 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2694 中支持手动压缩时的额外指令\n* fix：由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2807 中将网站构建工具从 npm 更改为 pnpm\n* perf(chat)：由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2823 中提升控制台聊天的重连兼容性\n* feat(console,backend)：由 @hikariming 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2695 中支持代理的持久化拖拽排序功能\n* fix(skills)：由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2775 中修复重命名技能时保留目录内容的问题\n* Fix WeChat channel：由 @xianzhisheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2597 中改进文件上传功能，并添加打字提示\n* Docs(skill)：由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2767 中更新技能文档\n* fix(qq)：由 @kobewl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2827 中在 WebSocket 会话恢复时重置重连状态\n* feat(parser)：由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2832 中为工具调用添加宽松解析功能\n* fix：由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2835 中修复版本相关问题\n* refactor(agents)：由 @jinliyl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2834 中修复命令处理器参数命名问题，并改进 Shell 工具的超时处理\n* docs(readme)：由 @zhijianma 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2836 中添加 Windows 平台的 WebView2 安装说明及 Web 认证详情\n* refactor(console)：由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2800 中使用 dayjs 的 relativeTime 方法实现 formatTimeAgo 的国际化\n\n## 新贡献者\n* @xianzhisheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2597 中完成了首次贡献\n* @kobewl 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2827 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0.post3...v1.0.1-beta.1","2026-04-02T12:51:19",{"id":218,"version":219,"summary_zh":220,"released_at":221},255340,"v1.0.0.post3","## 变更内容\n* chore(版本): 版本升级至 1.0.0p3，由 @xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2738 中完成\n* style: 技能与技能池及暗色风格，由 @zhaozhuang521 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2714 中完成\n* feat (ci): 增加触发 OSS 上传的选项，由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2740 中完成\n* fix(渠道)\u002F钉钉白名单，由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2718 中完成\n* 网站样式，由 @yuluo1007 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2722 中完成\n* feat(渠道): 添加 _on_process_completed 钩子，以支持飞书工作区路径中的 DONE 反馈，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2727 中完成\n* feat(渠道): 统一工作区跟踪器和旧路径的无文本防抖处理，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2724 中完成\n* chore: 移除 `AGENTS.md` 文件，由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2745 中完成\n* feat(技能): 批量删除技能、广播及下载功能，由 @Leirunlin 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2743 中完成\n* feat: 当代理被禁用时停止服务，由 @rayrayraykk 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2746 中完成\n* feat(控制台): 添加系统选项以切换暗色模式，由 @bowenliang123 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2678 中完成\n* feat(钉钉): 支持工作区跟踪器路径中的 AI 卡，并重构共享核心代码，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2741 中完成\n* Fix(提供者): CoPaw Local 默认使用 GPU && 修复探针镜像 && 优化本地提供者文档 && 修复 Windows 桌面模型下载问题，由 @pan-x-c 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2735 中完成\n* style: 修复技能名称，由 @xieyxclack 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2765 中完成\n* feat(脚本): 修复 exe 中缺少 copaw 命令的问题，由 @qbc2016 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2759 中完成\n* fix(wecom): 确保 Windows 守护进程中 wecom WS 线程的标准输入输出流正常运行，由 @hongxicheng 在 https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2760 中完成\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0.post2...v1.0.0.post3","2026-04-01T12:59:07",{"id":223,"version":224,"summary_zh":225,"released_at":226},255341,"v1.0.0.post2","## What's Changed\r\n* chore(version): bump version to 1.0.0.post2 by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2647\r\n* style(website): improve list marker visibility with softer color by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2648\r\n* Feat\u002Fconsole mcp by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2652\r\n* feat: avoid migrate skills every time when app start by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2649\r\n* fix(wecom): schedule reconnect in separate task on heartbeat failure by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2651\r\n* Website UI majorization by @yuluo1007 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2645\r\n* Doc(Provider): Improve FAQ for CoPaw-Flash deployment by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2661\r\n* fix: local by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2662\r\n* Refactoring(skill)\u002F Remove pool workspace sync for efficiency by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2659\r\n* style(tool): fix async tool status by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2676\r\n* refactor(console): reorder language options in dropdown by @bowenliang123 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2673\r\n* chore(deps): update reme-ai dependency to version 0.3.1.8 by @jinliyl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2654\r\n* Fix(Provider): Check repo before downloading and update CoPaw Flash doc by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2688\r\n* Perf\u002Foptimize skill list and refresh by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2687\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0.post1...v1.0.0.post2","2026-03-31T15:50:10",{"id":228,"version":229,"summary_zh":230,"released_at":231},255342,"v1.0.0.post1","## What's Changed\r\n* chore(version): bump version to 1.0.0.post1 by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2603\r\n* Feat\u002Fconsole message by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2604\r\n* fix: message by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2612\r\n* fix: prevent none output by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2608\r\n* fix(console): 修复暗黑模式下斜杠命令菜单的文字对比度 by @ltzu929 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2600\r\n* feat: update runtime to improve robustness by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2616\r\n* fix(doc): broken anchor links in README Table of Contents by @bowenliang123 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2614\r\n* fix: 支持 requires 字段的列表格式 by @Jacky51351684 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2504\r\n* fix: exception handling for migration by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2618\r\n* docs: Update comparison by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2626\r\n* fix(skill) signature cache & require by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2620\r\n* Fix(Provider): Llama.cpp windows using Nvidia-GPU, check macOS version before installation by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2625\r\n* fix(skill): req parsing to use safe formatter by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2630\r\n* fix(dingtalk): fallback to Open API when sessionWebhook expires for scheduled tasks by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2617\r\n* feat(chat): rename variable for clarity and adjust memory loading logic by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2638\r\n* fix: fix anyio version to avoid busy-wait loop by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2634\r\n* fix: tool guard when using thinking model by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2631\r\n* Revert \"fix(wecom): trigger reconnect on heartbeat failure to prevent permanent disconnection\" by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2641\r\n* feat(tool): fix \"\\n\" in shell command for windows  by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2635\r\n* Feat\u002Fconsole select agent by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2640\r\n* feat(model): support video analysis for multimodal models by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2627\r\n* chore(console): add type checking to format scripts by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2643\r\n\r\n## New Contributors\r\n* @Jacky51351684 made their first contribution in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2504\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0...v1.0.0.post1","2026-03-31T07:33:13",{"id":233,"version":234,"summary_zh":235,"released_at":236},255343,"v1.0.0","## ✨ Added\r\n\r\n**Multi-Agent System**\r\n\r\n- **Background Task Support**: Background execution for inter-agent communication with task tracking, status polling, and cancellation via CLI `--background` flag ([#2345](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2345))\r\n- **Agent Enable\u002FDisable Toggle**: Agents can be enabled or disabled through the console UI and API ([#2249](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2249))\r\n- **Unified Priority Queue System and `\u002Fstop` command**: Per-channel, per-session priority queue system with `\u002Fstop` command support for canceling running tasks and clearing queued messages ([#2411](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2411))\r\n\r\n**Providers and Models**\r\n\r\n- **CoPaw Local Model**: Built-in local model provider using llama.cpp with automatic download, configuration, and custom repository support ([#2419](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2419), [#2476](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2476))\r\n- **Scoped Active Model Selection**: Active model can be set globally in Settings or per-agent in Chat ([#2278](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2278))\r\n- **Global LLM Rate Limiter**: QPM (queries per minute) sliding window rate limiting, concurrency control with semaphore, and global 429 pause coordination with jitter to prevent thundering-herd retries ([#2282](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2282))\r\n\r\n**Security**\r\n\r\n- **System Reboot & Service Protection**: Tool guard rules to block system reboot, shutdown, service control, broad process kill, and privilege escalation commands ([#2333](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2333))\r\n- **Chinese Prompt Injection Detection**: Extended skill security scanner with Chinese regex patterns for prompt injection and jailbreak attempts ([#2381](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2381))\r\n\r\n**Console & UI**\r\n\r\n- **Download Page**: Download page for desktop installers with mirror sites ([#2555](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2555))\r\n- **Multimodal Preview**: Image, audio, video, and file attachments are displayed with proper preview URLs in history messages and streaming ([#2297](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2297), [#2332](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2332))\r\n- **Chat Session Labels**: Chat sessions display channel labels and icons for easy identification of message source ([#2483](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2483))\r\n- **Command Suggestions**: Slash-style command suggestions (`\u002Fclear`, `\u002Fcompact`, `\u002Fapprove`, `\u002Fdeny`) in the chat input ([#2415](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2415))\r\n\r\n**Channels**\r\n\r\n- **WeChat iLink Bot**: New WeChat personal channel using iLink Bot API ([#2260](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2260))\r\n- **Custom Channel HTTP Routes**: Custom channels can now register FastAPI routes for webhooks and custom endpoints ([#2140](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2140))\r\n- **Discord Bot Message Filtering**: New configuration option to control whether messages from other bots are processed ([#2122](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2122))\r\n- **DingTalk Widescreen Cards**: Support widescreen AI card layout in DingTalk ([#2238](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2238))\r\n- **WeCom Media Upload**: WebSocket-based media upload for improved reliability over HTTP download paths ([#2401](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2401))\r\n\r\n**Tools & Skills**\r\n\r\n- **Async Tool Execution**: Per-tool async execution flag with automatic registration of background task helpers (`view_task`, `wait_task`, `cancel_task`) ([#2391](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2391))\r\n- **Skill Pool Architecture**: Two-layer skill system with shared skill pool and per-agent workspace skills ([#2173](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2173), [#2436](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2436), [#2440](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2440), [#2477](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2477), [#2480](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2480))\r\n- **Browser CDP Support**: Chrome DevTools Protocol (CDP) integration for connecting to running Chrome instances, port scanning, cache clearing, and attaching to remote browsers ([#2294](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2294))\r\n- **Multi-Workspace Cookie Management**: Browser automation now uses per-workspace persistent user data directories for cookie isolation and persistence across restarts ([#2131](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2131))\r\n\r\n**Internationalization**\r\n\r\n- **Server-Side Language Persistence**: UI language preference is now persisted to server-side settings ([#2408](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2408))\r\n- **Expanded Multi-Language Support**: Additional console UI components now support multiple languages ([#2478](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2478), [#2508](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2508))\r\n\r\n## 🔄 Changed\r\n\r\n**Context a","2026-03-30T16:39:34",{"id":238,"version":239,"summary_zh":240,"released_at":241},255344,"v1.0.0-beta.3","## What's Changed\r\n* chore(version): bumping version to 1.0.0b3 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2530\r\n* chore(deps): update @agentscope-ai\u002Fchat dependency by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2531\r\n* fix(toolguard): fix frontend response by @gnipping in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2524\r\n* chore: add basic guideline for project structure and formating in AGENTS.md by @bowenliang123 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2532\r\n* fix: set default active session item to current chat id by @sanfran1068 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2543\r\n* docs(cli): update docs and fix log path by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2538\r\n* style: icon & style by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2540\r\n* chore: update system prompt by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2542\r\n* docs(skill)\u002F update term & figure by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2545\r\n* docs: update community and contributing docs by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2546\r\n* fix(formatter): remove file:\u002F\u002F for media urls by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2544\r\n* feat(skillscanner): support skill pool scanning && reduce false positive by @gnipping in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2539\r\n* Feat\u002Fconsole styles by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2552\r\n* fix(skilpoll): fix display issues in skill pool detection by @gnipping in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2551\r\n* feat: add download page by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2555\r\n* Doc(provider): Update provider doc by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2541\r\n* fix(mcp): mcp edit, add hint for mcp edit by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2556\r\n* docs: update figures related to the console by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2561\r\n* feat(formatter): enhance anthropic chat formatter by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2560\r\n* docs(install): update install by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2562\r\n* docs: Update roadmap  by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2563\r\n* docs: update console fig in readme by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2564\r\n* fix(dingtalk): save conversation_id for cron task message delivery by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2566\r\n* feat(wecom): compress oversized images before upload by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2568\r\n* docs: add QR Code for RedNote by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2570\r\n* chore(deps): update @agentscope-ai\u002Fchat to version 1.1.56 by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2571\r\n* Website UI by @yuluo1007 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2558\r\n* docs: update link by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2572\r\n* Fix(Provider): Update llama.cpp download link by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2567\r\n* chore(release): add release note for v1.0.0 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2573\r\n* feat(runner): prepend summary memory retrieval for chat history by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2574\r\n* Doc(Provider): Remove Stale Local Provider Configuration by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2577\r\n* add google tag to track copaw website traffic by @Osier-Yi in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2581\r\n* docs: fix homepage by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2575\r\n* update website by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2582\r\n* chore: update release note by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2583\r\n\r\n## New Contributors\r\n* @yuluo1007 made their first contribution in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2558\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0-beta.2...v1.0.0-beta.3","2026-03-30T15:59:14",{"id":243,"version":244,"summary_zh":245,"released_at":246},255345,"v1.0.0-beta.2","## What's Changed\r\n* chore(version): bumping version to 1.0.0b2 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2526\r\n* style: fix styles by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2522\r\n* refactor(react_agent): update recent max bytes configuration path by @jinliyl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2525\r\n* fix: lintfix by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2528\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv1.0.0-beta.1...v1.0.0-beta.2","2026-03-30T06:12:51",{"id":248,"version":249,"summary_zh":250,"released_at":251},255346,"v1.0.0-beta.1","## What's Changed\r\n* docs: update release note for v0.2.0 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2212\r\n* feat(tool): parse toll call from thinking and text content by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2192\r\n* chore(version): update version to 0.2.0.post1 by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2242\r\n* docs: update roadmap after v0.2.0 release by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2234\r\n* Feat\u002Fconsole chats by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2228\r\n* fix(cli): 修正本地回环地址的持久化处理 by @qzcpl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2241\r\n* feat(dingtalk): add card_auto_layout config for widescreen AI card by @wwx814 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2238\r\n* Feature\u002Fbrowser multi workspace cookie management by @x1n95c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2131\r\n* feat(agent): add enable\u002Fdisable toggle for agent management by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2249\r\n* fix(QQ): use AMR-specific ffmpeg params for QQ voice message conversion by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2248\r\n* fix(discord): prevent duplicate message processing on WebSocket reconnect by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2253\r\n* fix(feishu): prevent cross-workspace message misrouting in multi-instance setup by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2244\r\n* fix(console): fix upload file type in console by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2266\r\n* fear(heartbeat): support cron expressions in heartbeat by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2261\r\n* feat: update star hint for README & welcome bot by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2272\r\n* fix(cron): Modify the optional\u002Frequired fields for cron job by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2276\r\n* style: Style optimization by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2271\r\n* fix(test) avoid overwrite workspace by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2286\r\n* fix(telegram): prevent premature timeout by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2280\r\n* Fix: Multi-Agent Model Configuration and Multi-Tab Isolation by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2283\r\n* doc: add multi agent docs by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2290\r\n* feat(chat): enhance file handling with preview URLs and token support by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2297\r\n* [doc]Add update note for user that install from source by @garyzhang99 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2310\r\n* Feat(Provider): Refactor active model selection with scoped global and agent semantics by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2278\r\n* fix(agents): trim space before and after the agents workspace path by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2319\r\n* Feat(tool): browser CDP connect and cache clear by @x1n95c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2294\r\n* fix(skills): add version recommendation for himalaya skill by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2334\r\n* feat(security): Add protection for system reboot and service restart by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2333\r\n* feat(model): add global LLM rate limiter with concurrency control and 429 coordination by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2282\r\n* refactor(context): restructure copaw context management v2.0 and configuration by @jinliyl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2300\r\n* chore: bumping version to 1.0.0b1 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2358\r\n* feat: fix worker to 1 by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2360\r\n* feat: Add Background Task Support for Multi-Agent Collaboration by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2345\r\n* fix(feishu): add WebSocket reconnection, stale message filtering, and silent-disconnect detection by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2311\r\n* feat(discord): Add option to filter bot messages by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2122\r\n* feat(channel): add WeChat iLink Bot channel by @livehl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2260\r\n* chore(tools): update docstring in execute_shell_command by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2380\r\n* feat(console): support multimodal  preview in history  messages  and streaming by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2332\r\n* fix(feishu): correct stale message filtering with server clock offset by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2376\r\n* feat(channels): allow custom channels to register HTTP routes (WeChat ClawBot integrated) by @carlos999","2026-03-30T04:35:28",{"id":253,"version":254,"summary_zh":255,"released_at":256},255347,"v0.2.0.post1","## What's Changed\r\n* docs: update release note for v0.2.0 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2212\r\n* feat(tool): parse toll call from thinking and text content by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2192\r\n* chore(version): update version to 0.2.0.post1 by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2242\r\n* docs: update roadmap after v0.2.0 release by @cuiyuebing in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2234\r\n* Feat\u002Fconsole chats by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2228\r\n* fix(cli): 修正本地回环地址的持久化处理 by @qzcpl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2241\r\n* feat(dingtalk): add card_auto_layout config for widescreen AI card by @wwx814 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2238\r\n* Feature\u002Fbrowser multi workspace cookie management by @x1n95c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2131\r\n* feat(agent): add enable\u002Fdisable toggle for agent management by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2249\r\n* fix(QQ): use AMR-specific ffmpeg params for QQ voice message conversion by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2248\r\n* fix(discord): prevent duplicate message processing on WebSocket reconnect by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2253\r\n* fix(feishu): prevent cross-workspace message misrouting in multi-instance setup by @hongxicheng in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2244\r\n* fix(console): fix upload file type in console by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2266\r\n* fear(heartbeat): support cron expressions in heartbeat by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2261\r\n* feat: update star hint for README & welcome bot by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2272\r\n* fix(cron): Modify the optional\u002Frequired fields for cron job by @lalaliat in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2276\r\n* style: Style optimization by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2271\r\n* fix(test) avoid overwrite workspace by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2286\r\n* fix(telegram): prevent premature timeout by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2280\r\n* Fix: Multi-Agent Model Configuration and Multi-Tab Isolation by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2283\r\n* doc: add multi agent docs by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2290\r\n\r\n## New Contributors\r\n* @qzcpl made their first contribution in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2241\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv0.2.0...v0.2.0.post1","2026-03-25T13:02:39",{"id":258,"version":259,"summary_zh":260,"released_at":261},255348,"v0.2.0","## ✨ Added\r\n\r\n**Agent**\r\n\r\n- **Inter-Agent Communication**: Added `copaw agents` and `copaw message` CLI commands for listing agents, pushing messages to channels, and sending requests between agents ([#1959](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1959), [#2161](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2161))\r\n- **Built-in QA Agent**: Ships a pre-configured QA agent for answering CoPaw installation and usage questions ([#2189](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2189))\r\n- **Configurable LLM Auto-Retry**: LLM retry behavior is now configurable per agent from the console Settings page. ([#2002](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2002))\r\n- **Summarization Improvements**: When the agent reaches max iterations and enters summarization, a \"round ended\" notice is appended to guide the user ([#1965](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1965), [#2163](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2163))\r\n- **Config Auto-Repair**: If `config.json` is corrupted or has minor syntax errors, it is automatically repaired on load. Unrecoverable files are backed up with a unique suffix and CoPaw starts with default settings instead of crashing ([#2164](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2164))\r\n\r\n**Security**\r\n\r\n- **File Access Guard**: Added a configurable deny list for sensitive files and directories. When enabled, the agent's tools are blocked from reading or writing those paths ([#1762](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1762))\r\n- **Tool Guard Enhanced**: Tool guard is now enhanced and works correctly when multiple tools run in parallel ([#1984](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1984), [#1948](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1948))\r\n\r\n**Channels**\r\n\r\n- **Feishu \u002F Lark SDK Migration**: Migrated the Feishu channel to the official `lark-oapi` SDK with native async calls, and added a region selector (Feishu China vs Lark International). ([#1946](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1946), [#2108](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2108), [#2178](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2178))\r\n- **XiaoYi File & Image Support**: Users can now send images and documents through the XiaoYi channel. ([#1885](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1885))\r\n\r\n**Console & UI**\r\n\r\n- **Audio, Video & Speech Input**: Console chat now supports sending and displaying audio, video and speech attachments natively ([#2114](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2114), [#2180](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2180))\r\n- **Stream Reconnection**: If a page refresh occurs while the agent is still responding, the console automatically reconnects to the in-flight streaming session and preserves the user's input text ([#1931](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1931))\r\n- **Web Account Management**: Users can now change their username and password from the console sidebar when authentication is enabled ([#1989](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1989))\r\n- **Model Provider Search**: Added a search box on the model provider settings page to filter providers by name ([#2062](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2062))\r\n- **Chat Scrollbar**: Added a styled scrollbar to the chat conversation area for easier navigation of long threads ([#2083](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2083))\r\n\r\n**Providers**\r\n\r\n- **Multimodal Capability Probing**: Models can now be probed for image and video support, and system prompt includes hints about the active model's multimodal capabilities ([#2045](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2045), [#2183](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2183))\r\n- **Gemini Tool Image Support**: Gemini models now correctly surface images returned by tools in the conversation, matching existing OpenAI behavior ([#2127](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2127))\r\n\r\n**Skills & Tools**\r\n\r\n- **Enhanced Grep & Glob Search**: The built-in grep and glob tools now skip heavy directories, enforce output size and time limits with configurable timeouts ([#2186](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2186))\r\n- **Workspace-Relative Tool Output**: Screenshots, PDFs, browser snapshots, and other tool output files are now saved under the agent's workspace directory instead of the process working directory ([#2159](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2159))\r\n\r\n## 🔄 Changed\r\n\r\n**Core & Lifecycle**\r\n\r\n- **Stable Prompts for KV Cache**: The environment context injected into each request now includes only the current date instead of a full timestamp, so prompt content stays stable across requests and improves LLM KV cache hit rates ([#2176](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2176))\r\n- **Faster CLI Startup**: CLI commands are now lazy-loaded so `copaw --help` and subcommands start significantly faster ([#2135](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F2135))\r\n\r\n**Channels**\r\n\r\n- **QQ Channel Refactor**: Simplified the QQ channel implementation and added comprehensive unit tests","2026-03-24T14:58:59",{"id":263,"version":264,"summary_zh":265,"released_at":266},255349,"v0.1.0.post1","## What's Changed\r\n* chore(version): bumping version to 0.1.1b1 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1882\r\n* fix: add HTTP 529 to retryable status codes for Anthropic overloaded errors by @ixiadao in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1891\r\n* fix(token_usage): replace threading.Lock with asyncio.Lock in record() by @mvanhorn in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1893\r\n* feat(channel): add file and image support for XiaoYi channel by @wwx814 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1885\r\n* feat: fix mac desktop release when dependency conflict by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1914\r\n* chore(deps): update reme-ai dependency to version 0.3.1.1 by @jinliyl in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1886\r\n* fix: 修复 Swagger 文档路由被 SPA 通配符覆盖的问题 by @mautops in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1847\r\n* Fix(Provider): Aligning the message format used for model connection testing with that used in actual usage. by @pan-x-c in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1909\r\n* fix(cron): handle CancelledError so cancelled jobs report correct status by @mvanhorn in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1894\r\n* style(console): select by @zhaozhuang521 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1920\r\n* perf: Fix console static directory resolution for non-repo CWD launches by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1923\r\n* fix(win): fix shell command on win by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1919\r\n* perf(ci): optimize test workflow to reduce resource consumption by @rayrayraykk in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1940\r\n* fix(migration) media\u002Femb_cache\u002F.md by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1922\r\n* feat: loosen \u002Fapprove matching by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1948\r\n* fix (channel): remove local file path whitelist by @Leirunlin in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1947\r\n* fix(mcp): skip failed mcp by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1956\r\n* refactor(channels): simplify QQ channel implementation and add unit tests by @leoleils in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1930\r\n* feat(api): implement header building and filename generation for work… by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1953\r\n* chore(config): update load config by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1939\r\n* refactor(file_handling): replace aiofiles with synchronous file opera… by @zhijianma in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1962\r\n* chore: bumping version to 0.1.0p1 by @xieyxclack in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1964\r\n* feat(summarizing): filter tool use blocks and add comment by @qbc2016 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1965\r\n* docker image migration by @yuanxs21 in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1949\r\n\r\n## New Contributors\r\n* @ixiadao made their first contribution in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1891\r\n* @leoleils made their first contribution in https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1930\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fcompare\u002Fv0.1.0...v0.1.0.post1","2026-03-20T11:46:49",{"id":268,"version":269,"summary_zh":270,"released_at":271},255350,"v0.1.0","## ✨ Added\r\n\r\n**Architecture**\r\n\r\n- **Multi-Agent \u002F Multi-Workspace Architecture**: Support running multiple agents simultaneously, each with its own isolated workspace containing independent config, memory, skills, and tools. Includes a console agent selector for easy switching between agents ([#1375](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1375), [#1625](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1625), [#1670](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1670), [#1714](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1714), [#1664](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1664), [#1614](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1614))\r\n- **Context Management**: Added token counting, `\u002Fdump_history` and `\u002Fload_history` commands, configurable history length limits, and progress indicators during memory compaction ([#1628](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1628), [#1754](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1754), [#1689](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1689))\r\n\r\n**Security**\r\n\r\n- **Skill Security Scanner**: Skills are now scanned for security risks (prompt injection, command injection, hardcoded secrets, data exfiltration) before installation using static analysis ([#564](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F564))\r\n- **Destructive Shell Command Detection**: Added security rules to detect dangerous shell commands such as disk formatting, fork bombs, reverse shells, and privilege escalation attempts ([#1484](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1484), [#1529](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1529))\r\n- **Web Authentication**: Added optional web authentication with single-user registration, token-based login, localhost bypass, and CLI command ([#1329](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1329), [#1666](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1666), [#1719](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1719))\r\n\r\n**Channels**\r\n\r\n- **WeCom Channel**: Added WeCom as a messaging channel with media support, QR code access, console configuration UI ([#1407](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1407), [#1706](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1706), [#1725](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1725), [#1728](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1728), [#1681](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1681), [#1747](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1747), [#1766](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1766), [#1618](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1618), [#1631](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1631), [#1860](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1860), [#1856](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1856))\r\n- **XiaoYi Channel**: Added XiaoYi as a messaging channel ([#1213](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1213))\r\n- **DingTalk AI Card Reply**: DingTalk now supports AI Card-based replies with incremental streaming, falling back to webhook\u002Fmarkdown when unavailable ([#1118](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1118))\r\n\r\n**Skills & Tools**\r\n\r\n- **LobeHub Skill Import**: Skills can now be imported directly from LobeHub ([#1350](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1350))\r\n- **ModelScope Skill Hub**: Skills can now be imported from ModelScope Skill Hub ([#1673](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1673))\r\n- **Version-Aware Builtin Skill Sync**: Built-in skills now track versions and automatically sync updates while preserving user customizations ([#1674](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1674), [#1716](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1716), [#1749](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1749))\r\n- **Guidance Skill**: Added a built-in skill that answers CoPaw installation and configuration questions using local docs and the official website ([#1522](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1522))\r\n- **Zip Archive Skill Import**: Skills can now be imported by uploading zip archives directly from the console ([#1840](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1840))\r\n- **Built-in Tools**: Added `glob_search` and `grep_search` as built-in tools for file discovery and content search within the agent workspace ([#1852](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1852))\r\n- **MCP Header Environment Variables**: Support MCP HTTP client headers expanding `${ENV_VAR}` placeholders ([#1629](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1629))\r\n\r\n**Multimodal**\r\n\r\n- **Console Multimodal Chat**: The console chat now supports sending images and files alongside text messages ([#1772](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1772))\r\n- **View Image Tool**: Added a `view_image` tool that lets the LLM analyze local images for multimodal conversations ([#1526](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F1526))\r\n- **Non-Multimodal LLM Media Fallback**: When a non-multimodal LLM receives image\u002Faudio\u002Fvideo content, the system now automatically strips media blocks and retries the request ([#1676](https:\u002F\u002Fgithub.com\u002Fagentscope-ai\u002FCoPaw\u002Fpull\u002F","2026-03-19T14:38:18"]