[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-giselles-ai--giselle":3,"tool-giselles-ai--giselle":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",140436,2,"2026-04-05T23:32:43",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":79,"owner_twitter":80,"owner_website":81,"owner_url":82,"languages":83,"stars":104,"forks":105,"last_commit_at":106,"license":107,"difficulty_score":23,"env_os":108,"env_gpu":108,"env_ram":108,"env_deps":109,"category_tags":116,"github_topics":117,"view_count":10,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":137,"updated_at":138,"faqs":139,"releases":169},3355,"giselles-ai\u002Fgiselle","giselle","Giselle: AI App Builder. Open Source.","Giselle 是一款开源的 AI 智能体工作流构建平台，旨在打造无缝的人机协作体验。它主要解决了传统 AI 应用开发门槛高、流程复杂的问题，让用户无需编写大量代码，即可通过可视化界面轻松设计、编排并运行复杂的 AI 自动化任务。\n\n无论是希望快速验证想法的开发者、需要定制自动化流程的研究人员，还是想要探索 AI 潜力的产品设计师，都能利用 Giselle 高效地构建专属智能体。其核心亮点在于强大的“可视化智能体编辑器”，支持拖拽式操作来组合多个大模型（如 OpenAI、Anthropic、Google AI），并内置了知识库存储与团队协作功能。这意味着用户不仅能灵活调度不同模型的能力，还能让智能体基于私有数据做出更精准的决策。作为一个由社区驱动的开源项目，Giselle 提供了透明的代码基础和灵活的本地部署方案，帮助用户在保护数据隐私的同时，加速从概念到产品的交付过程。","\u003Cdiv align=\"center\">\n  \n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fgiselles-ai\u002Fgiselle.svg?style=social&label=Star\" alt=\"GitHub stars\" style=\"margin-right: 5px;\">\u003C\u002Fa>  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache2.0-blue.svg\" alt=\"License\" style=\"margin-right: 5px;\">\n  \u003Ca href=\"CONTRIBUTING.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPRs-welcome-brightgreen.svg\" alt=\"PRs Welcome\">\u003C\u002Fa>\n\n  \n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_e690b61e0184.png\" alt=\"Giselle logo\" height=\"100\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_3b97d58399c9.png\" alt=\"Giselle logo\" height=\"100\">\n\n  \n  \u003Cspan style=\"font-size: 18px; color: #666; margin-left: 15px;\">the AI agent studio powering product delivery\u003C\u002Fspan>\n\n\n  \u003Cp>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - Build and run AI workflows. Open source. | Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ftop-post-badge.svg?post_id=924550&theme=light&period=daily&t=1767082208449\">\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-featured&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - Build and run AI workflows. Open source. | Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ffeatured.svg?post_id=924550&theme=light&t=1767667659741\">\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - Build and run AI workflows. Open source. | Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ftop-post-badge.svg?post_id=924550&theme=light&period=weekly&t=1767667659741\">\u003C\u002Fa>\n  \u003C\u002Fp>\n\n  \u003C!-- Demo Video -->\n \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_233249c0b494.gif\" alt=\"Giselle Video\" width=\"100%\"> \u003Cbr\u002F>\n\n▶︎▶︎ [The YouTube video with audio can be found here](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=g1siFWk0GNs) ◀︎◀︎\n\n\n  \u003C!-- Light\u002FDark Mode GIFs -->\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_af84b7b7cc34.png\" alt=\"Giselle Demo\" width=\"400\"> \u003Cbr\u002F>\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_02b1418ab8a8.png\" alt=\"Giselle Demo\" width=\"400\">\n\n\u003C\u002Fdiv>\n\n## 👋 Introduction\n\nGiselle is an open source AI for agentic workflows, enabling seamless human-AI collaboration.\n\n\n\n## ⚡ Quick Start\n\nGet Giselle running locally in under 2 minutes:\n\n```bash\n# Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle.git\ncd giselle\n\n# Install dependencies\npnpm install\n\n# Create environment file\ntouch .env.local\n\n# Add your API key (at least one required)\necho 'OPENAI_API_KEY=\"your_openai_api_key_here\"' >> .env.local\n\n# Start development server\npnpm turbo dev\n```\n\nOpen [http:\u002F\u002Flocalhost:3000](http:\u002F\u002Flocalhost:3000) and start building your AI agents!\n\n> **Note**: You need at least one AI provider API key. Supported providers: OpenAI, Anthropic, Google AI.\n\n## ✨ Features\n\n\u003Cdiv align=\"center\">\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_51fc0aba0e49.png\" width=\"100\" alt=\"GitHub AI Operations\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_1c63eee67e4c.png\" width=\"100\" alt=\"Visual Agent Builder\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_41cf988a9ac1.png\" width=\"100\" alt=\"Multi-Model Composition\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_6911b3ea6e70.png\" width=\"100\" alt=\"Knowledge Store\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_de5ea21846b7.png\" width=\"100\" alt=\"Team Collaboration\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_23cd5f53e7aa.png\" width=\"100\" alt=\"Template Hub\">\n\n\u003C\u002Fdiv>\n\n- **⚡ GitHub AI Operations** - Automates issues, PRs, and deployments with AI\n- **🎨 Visual Agent Builder** - Create and modify agents in minutes using an intuitive drag-and-drop interface\n- **🤖 Multi-Model Composition** - Leverage GPT, Claude, Gemini, and more—agents select the best model for each task\n- **📁 Knowledge Store** - Access and search your code and data from one place. GitHub vector store integration supported\n- **👥 Team Collaboration** - Design agents collaboratively with shared configurations and contextual awareness *(In Development)*\n- **🚀 Template Hub** - Kickstart projects with one-click agent templates—contributed by the community *(In Development)*\n\n## 🎯 Use Cases\n\n- **📚 Research Assistant** - Automatically gather information from web and internal docs\n- **🔍 Code Reviewer** - AI-powered code review that integrates with your GitHub workflow  \n- **📄 Document Generator** - Auto-create PRDs, specs, and release notes from your codebase\n- **🔄 Workflow Automator** - Chain multiple AI models to handle complex business processes\n\n## 🚀 Using Giselle\n\n### ☁️ Cloud\n\nWe host [Giselle](https:\u002F\u002Fgiselles.ai\u002F) as a cloud service for anyone to use instantly. It has all the same features as the self-hosted version, and includes 30 minutes of free Agent time per month in the free plan.\n\n### 🏠 Self-hosting\n\nFollow this [starter guide](CONTRIBUTING.md#development-environment-setup) to get Giselle running in your environment.\n\n### 🎵 Vibe Coding Guide\n\nIf you're using AI coding assistants like Claude, Cursor, or WindSurf to help build with Giselle, check out our [Vibe Coding Guide](\u002Fdocs\u002Fvibe\u002F01-introduction.md). This guide explains:\n\n- What is vibe coding and how to approach it effectively\n- How to set up your Node.js environment and install dependencies\n- Understanding Giselle's project structure\n- Running the playground and connecting to LLM providers\n\nDesigned for both developers and non-engineers, this guide will help you harness the power of AI to build with Giselle without needing traditional coding expertise.\n\n## 🗺️ Roadmap\n\nGiselle is currently still in active development. The roadmap for the public repository is currently being created, and once it's finalized, we will update this README accordingly.\n\n## 🤝 Contributing\n\nYour contributions — big or small — help Giselle evolve and improve. Interested in joining us?\n\nHere's how you can contribute:\n\n- Star this repo ⭐\n- Follow us on social media: [Facebook](https:\u002F\u002Fwww.facebook.com\u002FGiselleAI\u002F), [X](https:\u002F\u002Fx.com\u002FGiselles_AI), [Instagram](https:\u002F\u002Fwww.instagram.com\u002Fgiselle_de_ai) and [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@Giselle_AI)\n- [Report a bug](https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002Fnew?template=1_bug_report.yml) you encounter while using Giselle\n- [Request a feature](https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fdiscussions\u002Fcategories\u002Fideas) you think would be helpful\n- [Submit a pull request](CONTRIBUTING.md#how-to-submit-a-pull-request) if you'd like to add new features or fix bugs\n\nFor more details, please see our [contributing guide](CONTRIBUTING.md).\n\n## 📄 License\n\nGiselle is licensed under the [Apache License Version 2.0](LICENSE).\n\nLicenses for third-party packages can be found in [docs\u002Fpackages-license.md](docs\u002Fpackages-license.md).\n","\u003Cdiv align=\"center\">\n  \n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fgiselles-ai\u002Fgiselle.svg?style=social&label=Star\" alt=\"GitHub stars\" style=\"margin-right: 5px;\">\u003C\u002Fa>  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache2.0-blue.svg\" alt=\"License\" style=\"margin-right: 5px;\">\n  \u003Ca href=\"CONTRIBUTING.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPRs-welcome-brightgreen.svg\" alt=\"PRs Welcome\">\u003C\u002Fa>\n\n  \n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_e690b61e0184.png\" alt=\"Giselle logo\" height=\"100\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_3b97d58399c9.png\" alt=\"Giselle logo\" height=\"100\">\n\n  \n  \u003Cspan style=\"font-size: 18px; color: #666; margin-left: 15px;\">赋能产品交付的AI智能体工作室\u003C\u002Fspan>\n\n\n  \u003Cp>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - 构建并运行AI工作流。开源。| Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ftop-post-badge.svg?post_id=924550&theme=light&period=daily&t=1767082208449\">\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-featured&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - 构建并运行AI工作流。开源。| Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ffeatured.svg?post_id=924550&theme=light&t=1767667659741\">\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fgiselle?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_campaign=badge-giselle\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg alt=\"Giselle - 构建并运行AI工作流。开源。| Product Hunt\" width=\"250\" height=\"54\" src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ftop-post-badge.svg?post_id=924550&theme=light&period=weekly&t=1767667659741\">\u003C\u002Fa>\n  \u003C\u002Fp>\n\n  \u003C!-- 演示视频 -->\n \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_233249c0b494.gif\" alt=\"Giselle视频\" width=\"100%\"> \u003Cbr\u002F>\n\n▶︎▶︎ [带有音频的YouTube视频请点击这里](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=g1siFWk0GNs) ◀︎◀︎\n\n\n  \u003C!-- 浅色\u002F深色模式动图 -->\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_af84b7b7cc34.png\" alt=\"Giselle演示\" width=\"400\"> \u003Cbr\u002F>\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_02b1418ab8a8.png\" alt=\"Giselle演示\" width=\"400\">\n\n\u003C\u002Fdiv>\n\n## 👋 简介\n\nGiselle是一款面向代理式工作流的开源AI工具，能够实现人与AI之间的无缝协作。\n\n\n\n## ⚡ 快速入门\n\n在不到2分钟内即可在本地运行Giselle：\n\n```bash\n# 克隆仓库\ngit clone https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle.git\ncd giselle\n\n# 安装依赖\npnpm install\n\n# 创建环境文件\ntouch .env.local\n\n# 添加你的API密钥（至少需要一个）\necho 'OPENAI_API_KEY=\"your_openai_api_key_here\"' >> .env.local\n\n# 启动开发服务器\npnpm turbo dev\n```\n\n打开[http:\u002F\u002Flocalhost:3000](http:\u002F\u002Flocalhost:3000)，开始构建你的AI智能体吧！\n\n> **注意**：你需要至少一个AI提供商的API密钥。支持的提供商包括：OpenAI、Anthropic、Google AI。\n\n## ✨ 功能\n\n\u003Cdiv align=\"center\">\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_51fc0aba0e49.png\" width=\"100\" alt=\"GitHub AI运维\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_1c63eee67e4c.png\" width=\"100\" alt=\"可视化智能体构建器\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_41cf988a9ac1.png\" width=\"100\" alt=\"多模型组合\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_6911b3ea6e70.png\" width=\"100\" alt=\"知识库\" style=\"margin-right: 25px;\">   \n\u003Cimg src \"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_de5ea21846b7.png\" width=\"100\" alt=\"团队协作\" style=\"margin-right: 25px;\">   \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_readme_23cd5f53e7aa.png\" width=\"100\" alt=\"模板中心\">\n\n\u003C\u002Fdiv>\n\n- **⚡ GitHub AI运维** - 使用AI自动化处理问题、拉取请求和部署\n- **🎨 可视化智能体构建器** - 通过直观的拖放界面，几分钟内即可创建和修改智能体\n- **🤖 多模型组合** - 利用GPT、Claude、Gemini等模型——智能体会为每项任务选择最佳模型\n- **📁 知识库** - 从一处访问和搜索代码及数据。支持GitHub向量存储集成\n- **👥 团队协作** - 通过共享配置和上下文感知功能，协同设计智能体 *(正在开发中)*\n- **🚀 模板中心** - 一键启动项目，使用社区贡献的智能体模板 *(正在开发中)*\n\n## 🎯 应用场景\n\n- **📚 研究助理** - 自动从网络和内部文档中收集信息\n- **🔍 代码审查员** - 集成到GitHub工作流中的AI驱动代码审查\n- **📄 文档生成器** - 根据代码库自动生成PRD、规格说明和发布说明\n- **🔄 工作流自动化** - 将多个AI模型串联起来，以处理复杂的业务流程\n\n## 🚀 使用Giselle\n\n### ☁️ 云端\n\n我们以云服务的形式托管了[Giselle](https:\u002F\u002Fgiselles.ai\u002F)，任何人都可以立即使用。它具备与自托管版本完全相同的功能，并且免费计划每月提供30分钟的智能体使用时间。\n\n### 🏠 自托管\n\n请按照这份[入门指南](CONTRIBUTING.md#development-environment-setup)来在你的环境中运行Giselle。\n\n### 🎵 Vibe编码指南\n\n如果你正在使用Claude、Cursor或WindSurf等AI编码助手来帮助你基于Giselle进行开发，请查看我们的[Vibe编码指南](\u002Fdocs\u002Fvibe\u002F01-introduction.md)。本指南将介绍：\n\n- 什么是vibe编码以及如何有效地进行操作\n- 如何设置Node.js环境并安装依赖\n- 理解Giselle的项目结构\n- 运行游乐场并与LLM提供商连接\n\n该指南专为开发者和非工程师设计，旨在帮助你无需传统编码技能即可利用AI的力量，基于Giselle进行开发。\n\n## 🗺️ 路线图\n\nGiselle目前仍在积极开发中。公共仓库的路线图正在制定中，一旦确定，我们将相应地更新此README文件。\n\n## 🤝 贡献\n\n无论大小，您的贡献都有助于 Giselle 的发展与改进。您有兴趣加入我们吗？\n\n以下是您可以参与贡献的方式：\n\n- 为本仓库点个赞 ⭐\n- 关注我们的社交媒体：[Facebook](https:\u002F\u002Fwww.facebook.com\u002FGiselleAI\u002F)、[X](https:\u002F\u002Fx.com\u002FGiselles_AI)、[Instagram](https:\u002F\u002Fwww.instagram.com\u002Fgiselle_de_ai) 和 [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@Giselle_AI)\n- [报告错误](https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002Fnew?template=1_bug_report.yml)，如果您在使用 Giselle 时遇到问题\n- [请求功能](https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fdiscussions\u002Fcategories\u002Fideas)，如果您认为某项功能会很有帮助\n- [提交拉取请求](CONTRIBUTING.md#how-to-submit-a-pull-request)，如果您希望添加新功能或修复 bug\n\n更多详情，请参阅我们的[贡献指南](CONTRIBUTING.md)。\n\n## 📄 许可证\n\nGiselle 采用 [Apache License Version 2.0](LICENSE) 许可证。\n\n第三方软件包的许可证信息可在 [docs\u002Fpackages-license.md](docs\u002Fpackages-license.md) 中找到。","# Giselle 快速上手指南\n\nGiselle 是一个开源的 AI 智能体（Agent）工作室，旨在通过可视化的拖拽界面和强大的多模型组合能力，实现流畅的人机协作与工作流自动化。\n\n## 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**：macOS, Linux 或 Windows (WSL2 推荐)\n*   **Node.js**：建议安装最新 LTS 版本 (v18+)\n*   **包管理器**：必须安装 [pnpm](https:\u002F\u002Fpnpm.io\u002F) (项目依赖管理工具)\n    ```bash\n    npm install -g pnpm\n    ```\n*   **API 密钥**：至少需要一个主流大模型提供商的 API Key，支持：\n    *   OpenAI\n    *   Anthropic (Claude)\n    *   Google AI (Gemini)\n\n> **提示**：国内开发者若遇到网络问题，建议在配置环境变量时通过代理访问相关 API 服务，或在 `pnpm` 配置中使用国内镜像源加速依赖下载。\n\n## 安装步骤\n\n只需几分钟即可在本地启动 Giselle：\n\n1.  **克隆仓库**\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle.git\n    cd giselle\n    ```\n\n2.  **安装依赖**\n    ```bash\n    pnpm install\n    ```\n\n3.  **配置环境变量**\n    创建本地环境文件并填入您的 API Key（以下以 OpenAI 为例，也可替换为 `ANTHROPIC_API_KEY` 或 `GOOGLE_GENERATIVE_AI_API_KEY`）：\n    ```bash\n    touch .env.local\n    echo 'OPENAI_API_KEY=\"your_openai_api_key_here\"' >> .env.local\n    ```\n\n4.  **启动开发服务器**\n    ```bash\n    pnpm turbo dev\n    ```\n\n启动成功后，终端会显示服务地址。\n\n## 基本使用\n\n1.  **访问界面**\n    打开浏览器访问 [http:\u002F\u002Flocalhost:3000](http:\u002F\u002Flocalhost:3000)。\n\n2.  **构建第一个智能体**\n    *   进入可视化编辑器界面。\n    *   从左侧组件栏拖拽节点（如 LLM 模型、知识库检索、代码执行等）到画布。\n    *   连接节点以定义工作流逻辑。\n    *   在右侧面板配置具体参数（如选择模型类型、输入 Prompt）。\n\n3.  **运行与测试**\n    点击界面上的“运行”按钮，即可实时查看 AI 智能体的执行结果。您可以尝试创建一个简单的“代码审查助手”或“文档生成器”工作流来体验其功能。\n\n---\n*更多高级用法（如 GitHub 集成、团队协作配置）请参考项目根目录下的 `CONTRIBUTING.md` 或官方 Vibe Coding 指南。*","某电商初创团队的产品经理急需构建一个能自动处理用户退货请求、查询库存并生成补偿方案的智能客服流程，但团队内缺乏专职后端开发人员。\n\n### 没有 giselle 时\n- **开发门槛高**：产品经理虽有清晰的业务逻辑，但因不懂代码，只能依赖工程师排期开发，导致需求积压数周无法落地。\n- **模型切换繁琐**：为了平衡成本与效果，需要人工编写大量胶水代码来串联 OpenAI 和 Anthropic 等不同模型，调试过程极易出错。\n- **知识更新滞后**：最新的退货政策和库存数据散落在不同文档中，难以实时同步给 AI，导致客服经常给出过时的错误答复。\n- **协作黑盒**：业务逻辑硬编码在脚本里，非技术人员无法查看或调整流程节点，每次微调都需要重新部署整个服务。\n\n### 使用 giselle 后\n- **可视化搭建**：产品经理直接在 giselle 的画布上通过拖拽节点，像画流程图一样构建了包含“意图识别 - 库存查询 - 方案生成”的完整代理工作流，当天即可上线。\n- **多模型无缝编排**：利用 giselle 的多模型组合功能，轻松设定由低成本模型处理常规问答，复杂争议自动路由给高性能模型，无需编写任何集成代码。\n- **知识库动态挂载**：将公司政策文档一键上传至 giselle 的知识存储区，AI 代理能实时检索最新数据，确保回复准确率达到 95% 以上。\n- **透明化迭代**：团队成员可在同一界面直观看到 Agent 的思考路径和执行步骤，业务人员能直接调整判断条件，实现了真正的“人机协同”迭代。\n\ngiselle 让非技术背景的业务专家也能独立构建并运维复杂的 AI 代理工作流，将产品想法到落地的周期从数周缩短至数小时。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgiselles-ai_giselle_e81da4a5.png","giselles-ai","Giselle","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fgiselles-ai_cd008160.png","AI for agentic workflows. Human-AI collaboration.",null,"Giselles_AI","https:\u002F\u002Fgiselles.ai","https:\u002F\u002Fgithub.com\u002Fgiselles-ai",[84,88,92,96,100],{"name":85,"color":86,"percentage":87},"TypeScript","#3178c6",98.3,{"name":89,"color":90,"percentage":91},"JavaScript","#f1e05a",0.8,{"name":93,"color":94,"percentage":95},"CSS","#663399",0.7,{"name":97,"color":98,"percentage":99},"PLpgSQL","#336790",0.2,{"name":101,"color":102,"percentage":103},"Handlebars","#f7931e",0,509,118,"2026-04-03T19:09:59","Apache-2.0","未说明",{"notes":110,"python":111,"dependencies":112},"该工具主要基于 Node.js 运行，使用 pnpm 作为包管理器，并通过 turbo 进行开发服务器启动。无需本地部署大型 AI 模型，但需要配置至少一个外部 AI 提供商的 API 密钥（支持 OpenAI、Anthropic 或 Google AI）。","未说明 (基于 Node.js 环境)",[113,114,115],"Node.js","pnpm","turbo",[15,14,13],[118,119,120,121,122,123,124,125,126,127,128,114,129,130,131,132,133,134,135,136],"agent","ai","ai-agent","genai","generative-ai","nextjs","no-code","openai","typescript","vercel","workflow","agent-builder","gemini","agentic-ai","anthropic","multiagent","rag","ai-agents","open-source","2026-03-27T02:49:30.150509","2026-04-06T08:46:19.182373",[140,145,150,155,160,164],{"id":141,"question_zh":142,"answer_zh":143,"source_url":144},15417,"如何在本地启动并运行 Playground 项目？","你可以参考项目中的设置文档来开始。相关代码位于 `internal-packages\u002Fworkflow-designer-ui\u002Fsrc\u002Feditor\u002Fproperties-panel\u002Ftext-generation-node-properties-panel\u002Findex.tsx`，完整的入门指南请查看：https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Ftree\u002F0868ecdef6cbcb856174b4b3cf571241ee07cd02\u002Fapps\u002Fplayground#getting-started","https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002F478",{"id":146,"question_zh":147,"answer_zh":148,"source_url":149},15418,"为什么上传的图片预览在刷新页面后无法显示？","这是因为 `FileAttachments` 组件需要 `basePath` 属性来构建远程图片 URL，但 `PlaygroundStageInput` 和 `TaskCompactStageInput` 组件未传递该属性。由于 `resolvedBasePath` 默认为空字符串，导致条件判断失败，服务器端图片 URL 无法生成。解决方法是在调用这些组件时显式传入正确的 `basePath` 属性。","https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002F2656",{"id":151,"question_zh":152,"answer_zh":153,"source_url":154},15419,"Markdown 渲染中如何支持超链接点击？","该问题已通过 PR #669 解决。主要方案是配置 `react-markdown` 库以正确渲染超链接，确保 Markdown 内容中的链接被转换为带有正确 `href` 属性的 HTML `\u003Ca>` 标签，使用户可以点击跳转。","https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002F469",{"id":156,"question_zh":157,"answer_zh":158,"source_url":159},15420,"代码库中存在拼写错误的目录名（如 depreacted），是否已修复？","是的，该问题已解决。原本拼写错误的目录名 `(depreacted)` 已被重命名为正确的 `(deprecated)`，同时相关的死代码（注释掉的导入语句）也被移除。","https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002F2383",{"id":161,"question_zh":162,"answer_zh":163,"source_url":144},15421,"当 Prompt 输入框为空时，如何防止用户误触生成按钮？","已在 Generator Node 中实现检查逻辑：当 Prompt 字段为空或仅包含空白字符时，禁用“Generate”按钮；当输入有效非空字符串时自动启用。此举可避免无效请求并提升用户体验，相关逻辑已集成到现有事件处理器中。",{"id":165,"question_zh":166,"answer_zh":167,"source_url":168},15422,"数据库迁移任务执行后如何验证结果？","迁移完成后，维护者已在生产环境进行验证并确认工作正常。具体检查包括确认相关列和 organizations 表已被成功删除。验证步骤包括在预演和生产环境中分别运行迁移脚本并检查数据库结构变化。","https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fissues\u002F131",[170,175,180,185,190,195,200,205,210,215,220,225,230,235,240,245,250,255,260,265],{"id":171,"version":172,"summary_zh":173,"released_at":174},90069,"v0.75.3","## 变更内容\n\n### 维护\n- chore(deps): 将 pnpm\u002Faction-setup 从 4.2.0 升级到 5.0.0 (#2852) @dependabot\n- chore(deps): 将 actions\u002Fcache 从 5.0.1 升级到 5.0.4 (#2853) @dependabot\n- chore(deps): 将 biomejs\u002Fsetup-biome 从 2.7.0 升级到 2.7.1 (#2854) @dependabot\n- chore(deps): 将 ruby\u002Fsetup-ruby 从 1.288.0 升级到 1.299.0 (#2855) @dependabot\n- chore(deps): 将 actions\u002Fsetup-node 从 6.0.0 升级到 6.3.0 (#2856) @dependabot\n- 移除自定义的 CodeQL 和依赖项审查工作流 (#2857) @shige\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.75.2...v0.75.3\n\n最初发布日期: 2026年4月1日","2026-04-03T00:08:22",{"id":176,"version":177,"summary_zh":178,"released_at":179},90070,"v0.75.2","## 变更内容\n\n### Bug 修复\n- 修复(studio): 需显式设置 SENTRY_ENABLE=true 才能激活 Sentry (#2849) @shige\n\n### 维护\n- 将 GitHub Actions 锁定到完整的提交 SHA (#2850) @shige\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.75.1...v0.75.2\n\n最初发布日期: 2026-03-30","2026-03-31T00:58:47",{"id":181,"version":182,"summary_zh":183,"released_at":184},90071,"v0.75.1","## 变更内容\n\n### 新功能\n- feat(workflow-designer-ui): 将专业版推荐模型更新至最新版本 (#2846) @WashizuRyo\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.75.0...v0.75.1\n\n最初发布日期: 2026年3月20日","2026-03-23T01:05:48",{"id":186,"version":187,"summary_zh":188,"released_at":189},90072,"v0.75.0","## 变更内容\n\n### 新功能\n- 功能（语言模型）：新增 Google\u002FGemini-3.1-Pro-Preview 模型 (#2839) @WashizuRyo\n- 功能（语言模型）：增加 OpenAI\u002FGPT-5.3-Codex 模型支持 (#2838) @WashizuRyo\n- 功能（语言模型）：新增 Google\u002FGemini-3.1-Flash-Image-Preview 支持 (#2840) @WashizuRyo\n\n### 错误修复\n- 修复（网络搜索）：使用 ipaddr.js 添加 SSRF 防护 (#2843) @WashizuRyo\n- 修复（语言模型）：在成本计算中移除模型 ID 中的提供商前缀 (#2845) @WashizuRyo\n\n### 依赖更新\n- 杂项（依赖）：将 npm_and_yarn 组中的 Next.js 版本从 16.1.5 升级至 16.1.7，涉及 1 个目录 (#2844) @dependabot\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.74.0...v0.75.0\n\n最初发布日期：2026年3月19日","2026-03-20T05:34:48",{"id":191,"version":192,"summary_zh":193,"released_at":194},90073,"v0.74.0","## 变更内容\n\n### 新功能\n- 功能（语言模型）：新增对 anthropic\u002Fclaude-sonnet-4.6 的支持 (#2836) @WashizuRyo\n- 功能（语言模型）：新增对 gpt-5.4 OpenAI 模型的支持 (#2837) @WashizuRyo\n\n### 问题修复\n- 修复（GitHub 工具）：在缺少 Webhook 密钥和签名时快速失败 (#2841) @WashizuRyo\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.73.0...v0.74.0\n\n最初发布日期：2026年3月18日","2026-03-19T03:17:41",{"id":196,"version":197,"summary_zh":198,"released_at":199},90074,"v0.73.0","## 变更内容\n\n### 新功能\n- feat: 从 API 密钥中移除 isInternalPlan 限制 (#2830) @WashizuRyo\n\n### 重构\n- refactor(react): 将数据存储的可用性从功能标志移至 DataStore 上下文 (#2832) @WashizuRyo\n\n### 依赖项\n- fix(deps): 更新 @giselles-ai\u002Fagent 至 v0.1.27 (#2831) @toyamarinyon\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.72.1...v0.73.0\n\n最初发布日期: 2026年3月17日","2026-03-18T01:10:29",{"id":201,"version":202,"summary_zh":203,"released_at":204},90075,"v0.72.1","## 变更内容\n\n### 杂项\n- chore: 移除结构化输出功能开关 (#2826) @WashizuRyo\n\n### 依赖\n- feat(deps): 更新 @giselles-ai\u002Fagent 至 v0.1.26 (#2827) @toyamarinyon\n\n最初发布日期：2026年3月13日","2026-03-16T00:19:18",{"id":206,"version":207,"summary_zh":208,"released_at":209},90076,"v0.72.0","## 变更内容\n\n### 新功能\n- feat(workflow-designer-ui): 验证结构化输出属性名 (#2824) @WashizuRyo\n- feat(text-editor): 支持在节点引用中使用结构化输出字段级别的提及 (#2820) @WashizuRyo\n\n### 依赖项\n- chore(studio): 将 @giselles-ai\u002Fagent 升级至 0.1.25 (#2823) @toyamarinyon\n- chore(deps): 在 1 个目录中批量升级 npm_and_yarn 组，包含 2 次更新 (#2822) @dependabot\n\n最初发布日期：2026年3月12日","2026-03-13T00:14:30",{"id":211,"version":212,"summary_zh":213,"released_at":214},90077,"v0.71.1","## 变更内容\n\n### 新功能\n- 功能（工作流设计器界面）：同步结束节点输出模式与源节点变更 (#2818) @WashizuRyo\n\n最初发布日期：2026年3月11日","2026-03-12T00:10:14",{"id":216,"version":217,"summary_zh":218,"released_at":219},90078,"v0.71.0","## 变更内容\n\n### 新功能\n- 功能（sdk）：在 runAndWait 中添加 Standard Schema v1 验证支持 (#2802) @WashizuRyo\n\n### Bug 修复\n- 修复（giselle）：将 RSC 序列化中的 `Object.create(null)` 替换为普通对象 (#2815) @WashizuRyo\n- 修复（structured-output）：从 End Node 结构化输出选择器中移除数组类型 (#2816) @WashizuRyo\n- 修复：恢复 execute-action 中的详尽类型检查 (#2814) @sputnik-mac\n\n### 依赖项\n- 功能（deps）：将 agent 依赖更新至 0.1.23 (#2813) @toyamarinyon\n- 将 @giselles-ai\u002Fagent 更新至 v0.1.24 (#2817) @toyamarinyon\n\n最初发布日期：2026年3月10日","2026-03-11T00:15:41",{"id":221,"version":222,"summary_zh":223,"released_at":224},90079,"v0.70.1","## What's Changed\n\n### Improvements\n- refactor(agent-api): consolidate routes into single catch-all handler using createAgentApi (#2812) @toyamarinyon\n\nOriginally released: 2026-03-09","2026-03-10T00:40:18",{"id":226,"version":227,"summary_zh":228,"released_at":229},90080,"v0.70.0","## What's Changed\n\n### New Features\n- feat: add AI-powered schema generation to structured output dialogs (#2795) @WashizuRyo\n- feat: update agent dependencies to new runtime and builder packages (#2806) @toyamarinyon\n- feat(deps): update agent-builder and agent-runtime to v0.1.17 (#2807) @toyamarinyon\n- feat(studio): update agent-builder and agent-runtime to v0.1.18 (#2810) @toyamarinyon\n\n### Bug Fixes\n- fix(deps): resolve webpack security vulnerabilities (#2804) @shige\n- fix(workflow-designer-ui): restore inline layout for text generation node output format (#2803) @WashizuRyo\n\nOriginally released: 2026-03-06","2026-03-09T00:06:39",{"id":231,"version":232,"summary_zh":233,"released_at":234},90081,"v0.69.0","## What's Changed\n\n### Features (Private Preview)\n- add build endpoint for centralized Vercel Sandbox snapshot creation (#2801) @toyamarinyon\n- feat(sdk): Add structured output support to SDK (#2776) @WashizuRyo\n- feat(studio): display structured output on task detail page (#2784) @WashizuRyo\n\n### Security\n- fix(security): bump tar from 7.5.8 to 7.5.10 (#2800) @shige\n- fix(security): bump hono to 4.12.4 and @hono\u002Fnode-server to 1.19.10 (#2799) @shige\n\n### Chore\n- chore(studio): update @giselles-ai\u002Fsandbox-agent to 0.1.15 (#2798) @toyamarinyon\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.68.0...v0.69.0\n\nOriginally released: 2026-03-05 (UTC)","2026-03-06T00:03:40",{"id":236,"version":237,"summary_zh":238,"released_at":239},90082,"v0.68.0","## What's Changed\n\n### Features (Private Preview)\n- feat(agent-api): add snapshot_id parameter to run route (#2794) @toyamarinyon\n- feat: Show API response example in Code tab and handle enum type in structured output (#2790) @WashizuRyo\n\n### Bug Fixes\n- fix(workflow-designer-ui): align text generation node output format layout with end node (#2789) @WashizuRyo\n- fix(giselle): enforce recursive coercion for nested object\u002Farray schemas (#2786) @WashizuRyo\n- fix(workflow-designer-ui): remove array item property expansion in output source picker (#2785) @WashizuRyo\n\n### Security\n- fix(deps): resolve qs arrayLimit bypass vulnerability (Dependabot #98) (#2791) @shige\n\n### Refactoring\n- refactor(workflow-designer-ui): remove clear button from output source picker (#2793) @WashizuRyo\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.67.3...v0.68.0\n\nOriginally released: 2026-03-04 (UTC)","2026-03-05T01:19:45",{"id":241,"version":242,"summary_zh":243,"released_at":244},90083,"v0.67.3","## What's Changed\n\n### Security\n- fix(deps): resolve serialize-javascript RCE vulnerability (Dependabot #109) (#2782) @shige\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.67.2...v0.67.3\n\nOriginally released: 2026-03-03 (UTC)","2026-03-04T00:11:19",{"id":246,"version":247,"summary_zh":248,"released_at":249},90084,"v0.67.2","## What's Changed\n\n### Features (Private Preview)\n- feat: add structured object output support for task API (#2769) @WashizuRyo\n\n### Security\n- fix(deps): resolve esbuild CORS vulnerability (#2780) @shige\n\n### Chores\n- chore(deps): bump ruby\u002Fsetup-ruby from 1.287.0 to 1.288.0 (#2779) @dependabot\n- chore(deps): bump actions\u002Fupload-artifact from 6 to 7 (#2778) @dependabot\n- chore(deps): bump changesets\u002Faction from 1.6.0 to 1.7.0 (#2777) @dependabot\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.67.1...v0.67.2\n\nOriginally released: 2026-03-02 (UTC)","2026-03-03T00:24:11",{"id":251,"version":252,"summary_zh":253,"released_at":254},90085,"v0.67.1","## What's Changed\n\n### Features (Private Preview)\n- feat(task): persist end node output format on task creation (#2765) @WashizuRyo\n\n### Bug Fixes\n- fix: use continue instead of return in handleEdgesChange (#2739) @apoorvdarshan\n\n### Security\n- fix(deps): bump js-yaml 3.14.1 → 3.14.2 to fix prototype pollution (#2766) @shige\n- fix(deps): remove minimatch override to fix ReDoS vulnerabilities (#2767) @shige\n\n### Chores\n- build: add SANDBOX_SNAPSHOT_ID environment variable to turbo config (#2768) @toyamarinyon\n- chore(studio): update @giselles-ai\u002Fsandbox-agent to 0.1.12 (#2771) @toyamarinyon\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.67.0...v0.67.1\n\nOriginally released: 2026-02-27 (UTC)","2026-03-02T00:20:11",{"id":256,"version":257,"summary_zh":258,"released_at":259},90086,"v0.67.0","## What's Changed\n\n### Features (Private Preview)\n- feat: add structured output form UI for End Node (#2753) @WashizuRyo\n- feat(agent-api): add support for Codex agent type (#2760) @toyamarinyon\n\n### Dependencies\n- chore: bump knip 5.67.1 → 5.85.0 (#2757) @shige\n- chore: bump rollup 4.49.0 → 4.59.0 (#2759) @shige\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.66.1...v0.67.0\n\nOriginally released: 2026-02-26 (UTC)","2026-02-27T00:16:30",{"id":261,"version":262,"summary_zh":263,"released_at":264},90087,"v0.66.1","## What's Changed\n\n### Features (Private Preview)\n- feat(protocol): add structured output protocol for End Node (#2750) @WashizuRyo\n\n### Dependencies\n- fix: patch babel runtime\u002Fhelpers transitive versions (#2752) @shige\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.66.0...v0.66.1\n\nOriginally released: 2026-02-25 (UTC)","2026-02-26T00:51:11",{"id":266,"version":267,"summary_zh":268,"released_at":269},90088,"v0.66.0","## What's Changed\n\n### Features (Private Preview)\n- feat: Add Agent API endpoints with authentication and rate limiting (#2745) @toyamarinyon\n\n### Security\n- chore(deps): bump ajv from 6.12.6 to 6.14.0 (CVE-2025-69873) (#2751) @shige\n\n### Chores\n- chore(config): add global environment variables and update next config (#2748) @toyamarinyon\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fgiselles-ai\u002Fgiselle\u002Fcompare\u002Fv0.65.1...v0.66.0\n\nOriginally released: 2026-02-24 (UTC)","2026-02-25T00:12:18"]