[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-archestra-ai--archestra":3,"tool-archestra-ai--archestra":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",151314,2,"2026-04-11T23:32:58",[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":32,"last_commit_at":59,"category_tags":60,"status":17},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[52,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":77,"owner_website":77,"owner_url":78,"languages":79,"stars":115,"forks":116,"last_commit_at":117,"license":118,"difficulty_score":10,"env_os":119,"env_gpu":120,"env_ram":120,"env_deps":121,"category_tags":127,"github_topics":128,"view_count":32,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":149,"updated_at":150,"faqs":151,"releases":179},6841,"archestra-ai\u002Farchestra","archestra","Enterprise AI Platform with guardrails, MCP registry, gateway & orchestrator","Archestra 是一款专为企业打造的原生 MCP（模型上下文协议）安全 AI 平台，旨在让组织内部的大规模 AI 应用变得简单、可控且安全。它通过提供统一的网关、编排器和注册表，解决了企业在引入 AI 时面临的工具管理混乱、数据泄露风险高以及成本不可控等核心痛点。\n\n过去，分散在各台机器上的 MCP 服务器难以统一管理，容易引发凭证滥用或敏感数据外泄。Archestra 将这些资源集中到安全的编排器中，支持在 Kubernetes 环境下运行，并内置了严格的“护栏”机制。其独特的技术亮点包括“双大模型”架构，利用独立的安全子代理隔离危险操作，有效防御提示词注入攻击；同时采用非概率性的安全策略，从根源上阻断数据非法流出，即便面对复杂的诱导指令也能确保企业数据安全。\n\n该平台非常适合企业的平台工程团队、开发者以及管理层使用。开发者可以无忧地构建和部署智能体，无需过度担忧底层安全问题；平台团队能够轻松治理全组织的 MCP 访问权限与凭证；而管理者则能获得清晰的 AI 使用全景视图，在推动全员（包括非技术人员）一键接入 AI 的同时，显著降低运营成本。Archestra 让企业既能享受 A","Archestra 是一款专为企业打造的原生 MCP（模型上下文协议）安全 AI 平台，旨在让组织内部的大规模 AI 应用变得简单、可控且安全。它通过提供统一的网关、编排器和注册表，解决了企业在引入 AI 时面临的工具管理混乱、数据泄露风险高以及成本不可控等核心痛点。\n\n过去，分散在各台机器上的 MCP 服务器难以统一管理，容易引发凭证滥用或敏感数据外泄。Archestra 将这些资源集中到安全的编排器中，支持在 Kubernetes 环境下运行，并内置了严格的“护栏”机制。其独特的技术亮点包括“双大模型”架构，利用独立的安全子代理隔离危险操作，有效防御提示词注入攻击；同时采用非概率性的安全策略，从根源上阻断数据非法流出，即便面对复杂的诱导指令也能确保企业数据安全。\n\n该平台非常适合企业的平台工程团队、开发者以及管理层使用。开发者可以无忧地构建和部署智能体，无需过度担忧底层安全问题；平台团队能够轻松治理全组织的 MCP 访问权限与凭证；而管理者则能获得清晰的 AI 使用全景视图，在推动全员（包括非技术人员）一键接入 AI 的同时，显著降低运营成本。Archestra 让企业既能享受 AI 带来的效率飞跃，又能牢牢守住安全底线。","# MCP-native Secure AI Platform\n\nSimplify AI usage in your company, providing user-friendly MCP toolbox, observability and control built on a strong security foundation.\n\n\u003Cdiv align=\"center\">\n\n[![License](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Farchestra-ai\u002Farchestra)](LICENSE)\n\u003Cimg alt=\"GitHub commit activity\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fm\u002Farchestra-ai\u002Farchestra\"\u002F>\n\u003Cimg alt=\"Github Last Commit\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002Farchestra-ai\u002Farchestra\"\u002F>\n[![Contributors](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Farchestra-ai\u002Farchestra)](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fgraphs\u002Fcontributors)\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fdocs\u002Fplatform-quickstart\">Getting Started\u003C\u002Fa>\n  - \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Freleases\">Releases\u003C\u002Fa>\n  - \u003Ca href=\"https:\u002F\u002Farchestra.ai\u002Fjoin-slack\">Slack Community\u003C\u002Fa>\n\u003C\u002Fp>\n\u003C\u002Fdiv>\n\n_For Platform teams:_\n\n- Mitigate MCP chaos, move MCP servers from individual machines to a centralized orchestrator\n- Manage how MCP access data and credentials usage\n- Mitigate data exfiltration risks\n- Manage AI costs\n- AI Observability\n\n_For Developers:_\n\n- Deploy your MCP servers org-wide\n- Build and deploy agents without worrying about security\n\n_For Management:_\n\n- Bring 1-click MCP adoption to the whole organization for technical and non-technical users\n- Reduce AI costs up to 96%\n- Get full visibility on AI adoption, usage and data access\n\n## 🚀 Quickstart with docker\n\n```\ndocker pull archestra\u002Fplatform:latest;\ndocker run -p 9000:9000 -p 3000:3000 \\\n  -e ARCHESTRA_QUICKSTART=true \\\n  -v \u002Fvar\u002Frun\u002Fdocker.sock:\u002Fvar\u002Frun\u002Fdocker.sock \\\n  -v archestra-postgres-data:\u002Fvar\u002Flib\u002Fpostgresql\u002Fdata \\\n  -v archestra-app-data:\u002Fapp\u002Fdata \\\n  archestra\u002Fplatform;\n```\n\n[Full Quickstart Guide →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-quickstart)\n\n## 👩‍💻 ChatGPT-like chat with MCPs\n\n🎁 with private company-wide prompt registry!\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_091d0857b7ee.png\" alt=\"ChatGPT-like chat\" \u002F>\n\u003C\u002Fdiv>\n\n## 📋 Private MCP registry with governance\n\nAdd MCPs to your private registry to share them with your team: self-hosted and remote, self-built and third-party.\n\n[Learn more about Private MCP Registry →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-private-registry)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_3f324b799c66.png\" alt=\"MCP Registry\" \u002F>\n\u003C\u002Fdiv>\n\n## ☁️ Kubernetes-native MCP orchestrator\n\nRun MCP servers in kubernetes, managing their state, API keys, OAuth.\n\n[Learn more about MCP Orchestrator →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-orchestrator)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_6f809a785bb3.png\" alt=\"MCP Orchestrator\" \u002F>\n\u003C\u002Fdiv>\n\n## 🤖 Security sub-agents\n\nIsolating dangerous tool responses from the main agent to prevent prompt injections.\n\n[Learn more about Dual LLM →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-dual-llm)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_4eccba63a616.png\" alt=\"Dual-LLM sub-agent\" \u002F>\n\u003C\u002Fdiv>\n\n## 🚫 Non-probabalistic security to prevent data exfiltration\n\nModels could consume prompt injections via MCP uncontrollably (read your inbox, read your GitHub issues, read your customer's inquiries) and follow them resulting in data exfiltration.\n\n[Learn more about Tool Guardrails →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-ai-tool-guardrails) | [The Lethal Trifecta →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-lethal-trifecta)\n\nLive demo of archestra security engine preventing data leak from the private GitHub repo to the public repo:\n[![Archestra Demo](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_cc3b550077a7.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=SkmluS-xzmM&t=2155s)\n\nRead more: [Simon Willison](https:\u002F\u002Fsimonwillison.net\u002F2025\u002FJun\u002F16\u002Fthe-lethal-trifecta\u002F), [The Economist](https:\u002F\u002Fwww.economist.com\u002Fleaders\u002F2025\u002F09\u002F25\u002Fhow-to-stop-ais-lethal-trifecta)\n\nExamples of hacks:\n[ChatGPT](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FApr\u002F14\u002Fnew-prompt-injection-attack-on-chatgpt-web-version-markdown-imag\u002F) (April 2023), [ChatGPT Plugins](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FMay\u002F19\u002Fchatgpt-prompt-injection\u002F) (May 2023), [Google Bard](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FNov\u002F4\u002Fhacking-google-bard-from-prompt-injection-to-data-exfiltration\u002F) (November 2023), [Writer.com](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FDec\u002F15\u002Fwritercom-indirect-prompt-injection\u002F) (December 2023), [Amazon Q](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FJan\u002F19\u002Faws-fixes-data-exfiltration\u002F) (January 2024), [Google NotebookLM](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FApr\u002F16\u002Fgoogle-notebooklm-data-exfiltration\u002F) (April 2024), [GitHub Copilot Chat](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FJun\u002F16\u002Fgithub-copilot-chat-prompt-injection\u002F) (June 2024), [Google AI Studio](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F7\u002Fgoogle-ai-studio-data-exfiltration-demo\u002F) (August 2024), [Microsoft Copilot](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F14\u002Fliving-off-microsoft-copilot\u002F) (August 2024), [Slack](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F20\u002Fdata-exfiltration-from-slack-ai\u002F) (August 2024), [Mistral Le Chat](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FOct\u002F22\u002Fimprompter\u002F) (October 2024), [xAI's Grok](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FDec\u002F16\u002Fsecurity-probllms-in-xais-grok\u002F) (December 2024), [Anthropic's Claude iOS app](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FDec\u002F17\u002Fjohann-rehberger\u002F) (December 2024), [ChatGPT Operator](https:\u002F\u002Fsimonwillison.net\u002F2025\u002FFeb\u002F17\u002Fchatgpt-operator-prompt-injection\u002F) (February 2025), [Notion 3.0](https:\u002F\u002Fwww.codeintegrity.ai\u002Fblog\u002Fnotion) (September 2024).\n\n## 💰 Cost monitoring, limits and dynamic optimization\n\nPer-team, per-agent or per-org cost monitoring and limitations. Dynamic optimizer allows to reduce cost up to 96% by simply switching to cheaper models automatically for simpler tasks.\n\n[Learn more about Costs & Limits →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-costs-and-limits)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_60bdeeed3b5a.png\" alt=\"Cost & Limits\" \u002F>\n\u003C\u002Fdiv>\n\n## 📊 Observability\n\nMetrics, traces and logs allowing to come to a conclusion about per-org, per-agent and per-team token and tool usage, and performance.\n\n[Learn more about Observability →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-observability)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_17c97bd09d5b.png\" alt=\"Observability\" \u002F>\n\u003C\u002Fdiv>\n\n## 👍 Ready for production\n\n1. ✅ Lightning fast, 45ms at 95p: [Performance & Latency benchmarks →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-performance-benchmarks)\n2. ✅ [Terraform provider →](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Fterraform-provider-archestra)\n3. ✅ [Helm Chart →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-deployment#helm-deployment-recommended-for-production)\n\n## 🤝 Contributing\n\nWe welcome contributions from the community!\n\n- [Contribution Guidelines →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fcontributing)\n- [Developer Quickstart →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-developer-quickstart)\n- [Security & Bug Bounty →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fsecurity)\n\nThank you for contributing and continuously making \u003Cb>Archestra\u003C\u002Fb> better, \u003Cb>you're awesome\u003C\u002Fb> 🫶\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_ad9cb6f93413.png\" \u002F>\n\u003C\u002Fa>\n\n---\n\n\u003Cdiv align=\"center\">\n  \u003Cbr \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fblog\u002Farchestra-joins-cncf-linux-foundation\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_df2980f43c28.png\" height=\"50\" alt=\"Linux Foundation\" \u002F>\u003C\u002Fa>\n        \n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fblog\u002Farchestra-joins-cncf-linux-foundation\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_fbcccac59307.png\" height=\"50\" alt=\"CNCF\" \u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n","# MCP原生安全AI平台\n\n简化贵公司的人工智能使用，提供用户友好的MCP工具箱、可观测性和控制功能，并建立在强大的安全基础之上。\n\n\u003Cdiv align=\"center\">\n\n[![许可证](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Farchestra-ai\u002Farchestra)](LICENSE)\n\u003Cimg alt=\"GitHub提交活动\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fm\u002Farchestra-ai\u002Farchestra\"\u002F>\n\u003Cimg alt=\"Github最后提交\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flast-commit\u002Farchestra-ai\u002Farchestra\"\u002F>\n[![贡献者](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Farchestra-ai\u002Farchestra)](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fgraphs\u002Fcontributors)\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fdocs\u002Fplatform-quickstart\">开始使用\u003C\u002Fa>\n  - \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Freleases\">发布版本\u003C\u002Fa>\n  - \u003Ca href=\"https:\u002F\u002Farchestra.ai\u002Fjoin-slack\">Slack社区\u003C\u002Fa>\n\u003C\u002Fp>\n\u003C\u002Fdiv>\n\n_面向平台团队：_\n\n- 消除MCP混乱局面，将MCP服务器从独立机器迁移到集中式编排器\n- 管理MCP访问数据和凭证的使用方式\n- 降低数据外泄风险\n- 控制AI成本\n- 实现AI可观测性\n\n_面向开发者：_\n\n- 在整个组织内部署您的MCP服务器\n- 构建并部署代理，无需担心安全性问题\n\n_面向管理层：_\n\n- 为技术和非技术用户提供一键式MCP采用方案，推动全组织范围内的AI普及\n- 将AI成本降低多达96%\n- 全面掌握AI的采用情况、使用情况以及数据访问权限\n\n## 🚀 使用Docker快速入门\n\n```\ndocker pull archestra\u002Fplatform:latest;\ndocker run -p 9000:9000 -p 3000:3000 \\\n  -e ARCHESTRA_QUICKSTART=true \\\n  -v \u002Fvar\u002Frun\u002Fdocker.sock:\u002Fvar\u002Frun\u002Fdocker.sock \\\n  -v archestra-postgres-data:\u002Fvar\u002Flib\u002Fpostgresql\u002Fdata \\\n  -v archestra-app-data:\u002Fapp\u002Fdata \\\n  archestra\u002Fplatform;\n```\n\n[完整快速入门指南 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-quickstart)\n\n## 👩‍💻 类似ChatGPT的MCP聊天功能\n\n🎁 配备私有企业级提示词注册表！\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_091d0857b7ee.png\" alt=\"类似ChatGPT的聊天\" \u002F>\n\u003C\u002Fdiv>\n\n## 📋 带有治理功能的私有MCP注册表\n\n将MCP添加到您的私有注册表中，以便与团队共享：无论是自托管还是远程部署，无论是自行构建还是第三方提供的模型。\n\n[了解更多关于私有MCP注册表的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-private-registry)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_3f324b799c66.png\" alt=\"MCP注册表\" \u002F>\n\u003C\u002Fdiv>\n\n## ☁️ Kubernetes原生MCP编排器\n\n在Kubernetes中运行MCP服务器，管理其状态、API密钥和OAuth认证。\n\n[了解更多关于MCP编排器的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-orchestrator)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_6f809a785bb3.png\" alt=\"MCP编排器\" \u002F>\n\u003C\u002Fdiv>\n\n## 🤖 安全子代理\n\n将危险工具的响应与主代理隔离，以防止提示注入攻击。\n\n[了解更多关于双LLM的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-dual-llm)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_4eccba63a616.png\" alt=\"Dual-LLM子代理\" \u002F>\n\u003C\u002Fdiv>\n\n## 🚫 非概率型安全机制，防止数据外泄\n\n模型可能会不受控制地通过MCP接收并执行恶意提示（例如读取您的收件箱、GitHub问题或客户咨询），从而导致数据外泄。\n\n[了解更多关于工具护栏的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-ai-tool-guardrails) | [致命三重奏 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-lethal-trifecta)\n\nArchestra安全引擎实时演示如何阻止数据从私有GitHub仓库泄露到公共仓库：\n[![Archestra演示](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_cc3b550077a7.jpg)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=SkmluS-xzmM&t=2155s)\n\n更多信息：[Simon Willison](https:\u002F\u002Fsimonwillison.net\u002F2025\u002FJun\u002F16\u002Fthe-lethal-trifecta\u002F)，[经济学人](https:\u002F\u002Fwww.economist.com\u002Fleaders\u002F2025\u002F09\u002F25\u002Fhow-to-stop-ais-lethal-trifecta)\n\n黑客案例：\n[ChatGPT](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FApr\u002F14\u002Fnew-prompt-injection-attack-on-chatgpt-web-version-markdown-imag\u002F)（2023年4月）、[ChatGPT插件](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FMay\u002F19\u002Fchatgpt-prompt-injection\u002F)（2023年5月）、[Google Bard](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FNov\u002F4\u002Fhacking-google-bard-from-prompt-injection-to-data-exfiltration\u002F)（2023年11月）、[Writer.com](https:\u002F\u002Fsimonwillison.net\u002F2023\u002FDec\u002F15\u002Fwritercom-indirect-prompt-injection\u002F)（2023年12月）、[Amazon Q](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FJan\u002F19\u002Faws-fixes-data-exfiltration\u002F)（2024年1月）、[Google NotebookLM](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FApr\u002F16\u002Fgoogle-notebooklm-data-exfiltration\u002F)（2024年4月）、[GitHub Copilot Chat](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FJun\u002F16\u002Fgithub-copilot-chat-prompt-injection\u002F)（2024年6月）、[Google AI Studio](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F7\u002Fgoogle-ai-studio-data-exfiltration-demo\u002F)（2024年8月）、[Microsoft Copilot](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F14\u002Fliving-off-microsoft-copilot\u002F)（2024年8月）、[Slack](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FAug\u002F20\u002Fdata-exfiltration-from-slack-ai\u002F)（2024年8月）、[Mistral Le Chat](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FOct\u002F22\u002Fimprompter\u002F)（2024年10月）、[xAI的Grok](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FDec\u002F16\u002Fsecurity-probllms-in-xais-grok\u002F)（2024年12月）、[Anthropic的Claude iOS应用](https:\u002F\u002Fsimonwillison.net\u002F2024\u002FDec\u002F17\u002Fjohann-rehberger\u002F)（2024年12月）、[ChatGPT Operator](https:\u002F\u002Fsimonwillison.net\u002F2025\u002FFeb\u002F17\u002Fchatgpt-operator-prompt-injection\u002F)（2025年2月）、[Notion 3.0](https:\u002F\u002Fwww.codeintegrity.ai\u002Fblog\u002Fnotion)（2024年9月）。\n\n## 💰 成本监控、限制与动态优化\n\n按团队、按代理或按组织进行成本监控和限制。动态优化器可通过自动切换到更便宜的模型来处理简单任务，从而将成本降低多达96%。\n\n[了解更多关于成本与限制的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-costs-and-limits)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_60bdeeed3b5a.png\" alt=\"成本与限制\" \u002F>\n\u003C\u002Fdiv>\n\n## 📊 可观测性\n\n通过指标、跟踪和日志，您可以了解整个组织、各个代理以及各团队的令牌和工具使用情况及性能表现。\n\n[了解更多关于可观测性的信息 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-observability)\n\n\u003Cdiv align=\"center\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_17c97bd09d5b.png\" alt=\"可观测性\" \u002F>\n\u003C\u002Fdiv>\n\n## 👍 已准备好投入生产\n\n1. ✅ 极速响应，95%情况下仅需45毫秒：[性能与延迟基准测试 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-performance-benchmarks)\n2. ✅ [Terraform提供商 →](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Fterraform-provider-archestra)\n3. ✅ [Helm图表 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-deployment#helm-deployment-recommended-for-production)\n\n## 🤝 贡献\n\n我们欢迎社区的贡献！\n\n- [贡献指南 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fcontributing)\n- [开发者快速入门 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fplatform-developer-quickstart)\n- [安全与漏洞奖励计划 →](https:\u002F\u002Farchestra.ai\u002Fdocs\u002Fsecurity)\n\n感谢您的贡献，让\u003Cb>Archestra\u003C\u002Fb> 不断变得更好，\u003Cb>您太棒了\u003C\u002Fb> 🫶\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_ad9cb6f93413.png\" \u002F>\n\u003C\u002Fa>\n\n---\n\n\u003Cdiv align=\"center\">\n  \u003Cbr \u002F>\n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fblog\u002Farchestra-joins-cncf-linux-foundation\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_df2980f43c28.png\" height=\"50\" alt=\"Linux 基金会\" \u002F>\u003C\u002Fa>\n        \n  \u003Ca href=\"https:\u002F\u002Fwww.archestra.ai\u002Fblog\u002Farchestra-joins-cncf-linux-foundation\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_readme_fbcccac59307.png\" height=\"50\" alt=\"CNCF\" \u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>","# Archestra 快速上手指南\n\nArchestra 是一个原生的 MCP（Model Context Protocol）安全 AI 平台，旨在帮助企业简化 AI 使用，提供用户友好的 MCP 工具箱、可观测性以及基于坚实安全基础的控制能力。\n\n## 环境准备\n\n在开始之前，请确保您的系统满足以下要求：\n\n*   **操作系统**：支持 Docker 的 Linux、macOS 或 Windows 系统。\n*   **核心依赖**：\n    *   [Docker](https:\u002F\u002Fwww.docker.com\u002F) (已安装并正在运行)\n    *   `docker-compose` (可选，本指南使用单条 docker 命令)\n*   **网络要求**：能够访问 Docker Hub 以拉取镜像。\n    *   *注：若在国内网络环境下拉取缓慢，建议配置 Docker 国内镜像加速器。*\n*   **端口占用**：确保宿主机的 `9000` 和 `3000` 端口未被占用。\n\n## 安装步骤\n\nArchestra 提供了基于 Docker 的快速启动方案，只需一条命令即可部署包含数据库和应用服务的完整平台。\n\n在终端中执行以下命令：\n\n```bash\ndocker pull archestra\u002Fplatform:latest;\ndocker run -p 9000:9000 -p 3000:3000 \\\n  -e ARCHESTRA_QUICKSTART=true \\\n  -v \u002Fvar\u002Frun\u002Fdocker.sock:\u002Fvar\u002Frun\u002Fdocker.sock \\\n  -v archestra-postgres-data:\u002Fvar\u002Flib\u002Fpostgresql\u002Fdata \\\n  -v archestra-app-data:\u002Fapp\u002Fdata \\\n  archestra\u002Fplatform;\n```\n\n**命令说明：**\n*   `-p 9000:9000 -p 3000:3000`：映射平台服务端口。\n*   `-e ARCHESTRA_QUICKSTART=true`：启用快速启动模式，自动初始化示例配置。\n*   `-v \u002Fvar\u002Frun\u002Fdocker.sock:...`：允许平台管理本地的 MCP 服务器容器。\n*   `-v ...`：持久化存储 PostgreSQL 数据和应用程序数据。\n\n等待容器启动完成后，即可通过浏览器访问。\n\n## 基本使用\n\n### 1. 访问平台界面\n打开浏览器，访问以下地址（根据部署环境调整 IP）：\n*   **主控制台**：`http:\u002F\u002Flocalhost:9000`\n*   **聊天界面**：`http:\u002F\u002Flocalhost:3000`\n\n### 2. 体验类 ChatGPT 的 MCP 聊天\nArchestra 内置了类似 ChatGPT 的聊天界面，并集成了公司级的私有提示词注册表。\n*   登录界面后，直接在对话框中输入指令。\n*   平台会自动调用已注册的 MCP 工具（如读取文件、查询数据库等），同时通过安全子代理（Security Sub-agents）防止提示词注入和数据泄露。\n\n### 3. 管理私有 MCP 注册表\n您可以将自建或第三方的 MCP 服务器添加到私有注册表中，供团队共享：\n*   进入 **MCP Registry** 页面。\n*   添加自托管或远程 MCP 服务器配置。\n*   平台将统一管理服务状态、API 密钥及 OAuth 认证（支持 Kubernetes 原生编排）。\n\n### 4. 监控成本与可观测性\n*   **成本控制**：在仪表盘查看按团队、代理或组织划分的 Token 消耗。开启动态优化器可自动为简单任务切换至更便宜的模型，最高降低 96% 成本。\n*   **可观测性**：查看详细的 Metrics、Traces 和 Logs，分析 Token 使用情况、工具调用性能及数据访问记录。\n\n---\n*更多高级功能（如 Terraform 部署、Helm Chart 生产环境部署）请参考官方文档。*","某中型金融科技公司的平台团队正试图将基于 MCP（模型上下文协议）的 AI 助手推广至全公司，以自动化处理客户工单和内部数据查询。\n\n### 没有 archestra 时\n- **部署混乱且难以扩展**：开发人员各自在本地机器上运行 MCP 服务器，缺乏统一编排，导致服务状态不稳定且无法在 Kubernetes 环境中集中管理。\n- **严重的数据泄露风险**：AI 代理容易受到提示词注入攻击，可能恶意读取员工邮箱、GitHub 私有代码库或客户敏感数据并外传至公共网络。\n- **成本与权限失控**：管理层无法监控各部门的 AI 调用量和 Token 消耗，导致账单激增，且无法精细控制不同角色对敏感数据工具的访问凭证。\n- **协作壁垒高**：非技术人员无法安全使用内部开发的 AI 工具，每次新工具上线都需要繁琐的安全审查和手动配置。\n\n### 使用 archestra 后\n- **集中化云原生编排**：通过 archestra 的 Kubernetes 原生编排器，团队将分散的 MCP 服务器统一迁移至集群，自动管理 API 密钥、OAuth 认证及服务状态，实现一键部署。\n- **构建非概率性安全防线**：利用 archestra 的“双大模型”架构和工具护栏，系统能自动隔离并拦截恶意的提示词注入，从根源上杜绝了从私有 repo 向公网泄露数据的风险。\n- **全景可观测性与成本优化**：管理层通过仪表盘实时掌握全组织的 AI 采用率、数据访问路径及用量，成功将无效的 AI 支出降低了 96%，并实现了细粒度的权限管控。\n- **全员安全自助服务**：借助私有 MCP 注册中心，业务人员也能安全地调用经审核的工具箱，无需担心底层安全配置，极大加速了 AI 在组织内的落地。\n\narchestra 不仅解决了企业级 AI 应用中的安全恐慌与运维混乱，更让组织在可控的成本下实现了安全、高效的智能化转型。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Farchestra-ai_archestra_091d0857.png","archestra-ai","Archestra","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Farchestra-ai_ba95c304.png","",null,"https:\u002F\u002Fgithub.com\u002Farchestra-ai",[80,84,88,92,96,99,102,105,109,112],{"name":81,"color":82,"percentage":83},"TypeScript","#3178c6",98.5,{"name":85,"color":86,"percentage":87},"CSS","#663399",0.7,{"name":89,"color":90,"percentage":91},"Shell","#89e051",0.3,{"name":93,"color":94,"percentage":95},"Python","#3572A5",0.1,{"name":97,"color":98,"percentage":95},"Dockerfile","#384d54",{"name":100,"color":101,"percentage":95},"Go Template","#00ADD8",{"name":103,"color":104,"percentage":95},"HTML","#e34c26",{"name":106,"color":107,"percentage":108},"Makefile","#427819",0,{"name":110,"color":111,"percentage":108},"Starlark","#76d275",{"name":113,"color":114,"percentage":108},"JavaScript","#f1e05a",3541,422,"2026-04-12T02:38:42","AGPL-3.0","Linux, macOS, Windows","未说明",{"notes":122,"python":120,"dependencies":123},"该工具主要通过 Docker 容器化部署（推荐生产环境使用 Helm Chart 或 Kubernetes）。快速启动需要安装 Docker 并挂载 Docker Socket (\u002Fvar\u002Frun\u002Fdocker.sock)。README 中未明确提及具体的操作系统限制、GPU 需求、内存大小或 Python 版本，因为核心逻辑封装在 Docker 镜像中。支持自托管和远程 MCP 服务器编排。",[124,125,126],"Docker","Kubernetes (可选)","PostgreSQL (内置)",[14,13,15,35,52],[129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148],"a2a","acp","agent","ai","mcp","mcp-client","mcp-server","runtime","mcp-host","mcp-servers","mcp-tools","claude","deepseek","gemini","openai","a2a-mcp","chatgpt","chatgpt-api","k8s","mcp-gateway","2026-03-27T02:49:30.150509","2026-04-12T16:43:16.118365",[152,157,162,167,171,175],{"id":153,"question_zh":154,"answer_zh":155,"source_url":156},30854,"如何参与 Archestra 项目的赏金任务（Bounty）并获取报酬？","参与赏金任务需遵循以下步骤：\n1. 在 Issue 下评论 `\u002Fattempt #Issue编号` 并附上你的实施计划。\n2. 等待核心团队成员将你分配（assign）到该 Issue。\n3. 开始工作前阅读贡献指南：https:\u002F\u002Fwww.archestra.ai\u002Fdocs\u002Fcontributing\n4. 提交工作时创建 Pull Request，并在 PR 正文中包含 `\u002Fclaim #Issue编号` 以声明赏金。\n5. 报酬支付：奖励发放后 2-5 天内接收 100% 赏金（需确保你所在的国家\u002F地区支持付款，详见 https:\u002F\u002Falgora.io\u002Fdocs\u002Fpayments#supported-countries-regions）。","https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F1446",{"id":158,"question_zh":159,"answer_zh":160,"source_url":161},30855,"在提交关于赏金任务的 PR 之前有什么注意事项？","在被正式分配（assigned）到该 Issue 之前，请勿发布 PR。正确的流程是先与团队沟通范围，获得分配后再开始工作并提交 PR。如果提前提交了 PR，可能需要关闭并重新按照流程进行，先对齐范围再行动。","https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F1301",{"id":163,"question_zh":164,"answer_zh":165,"source_url":166},30856,"如何在聊天 UI 中显示不同模型的能力（Capabilities）？","实现方案包括：\n1. 创建能力注册表：在前端配置文件中（如 `platform\u002Ffrontend\u002Fsrc\u002Fconfig\u002Fmodel-capabilities.ts`）映射模型 ID 与其能力（如视觉、函数调用等）。\n2. 定义 `ModelCapability` 类型。\n3. 针对不同 LLM 提供商，研究其 API 是否返回这些详细信息；若不返回，需提出获取数据的替代方案。\n4. 修改当前的模型选择器组件以展示这些能力信息。\n最终交付物需包含对上述问题的回答以及展示实现效果的视频演示。","https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F2041",{"id":168,"question_zh":169,"answer_zh":170,"source_url":156},30857,"目前是否支持通过 Terraform 管理自定义 RBAC 角色和用户分配？","截至最近更新，核心团队的优先级有所调整，Terraform Provider 的相关工作暂时被搁置（on the back-burner）。虽然有关于添加 `archestra_role` 和 `archestra_user_role_assignment` 资源的提议，但目前后端尚未暴露相关的角色管理或用户查找端点（OpenAPI 规范中暂无）。团队计划在不久的将来重新审视此功能，建议关注后续更新。",{"id":172,"question_zh":173,"answer_zh":174,"source_url":161},30858,"如何将 MCP Apps 集成到 Archestra Chat UI 中？","集成 MCP Apps 需要完成以下关键步骤：\n1. 在 Archestra Chat UI 中添加对 MCP Apps 的支持（参考 Model Context Protocol 规范）。\n2. 确保通过 MCP Gateway 在第三方 UI 中正常工作。\n3. 确保通过 LLM Gateway 在第三方 UI 中正常工作。\n4. 至少测试两个真实厂商的 MCP 实现（例如 n8n-mcp 和 excalidraw-mcp），并将它们编入目录。\n验收标准是除了 PR 代码外，还需提供一个展示上述四步均能正常运行的演示（Demo）。",{"id":176,"question_zh":177,"answer_zh":178,"source_url":161},30859,"Archestra 项目对贡献者的技术背景有什么要求？","根据赏金任务的反馈，理想的贡献者通常具备以下经验：\n- **TypeScript\u002FJavaScript**: 5 年以上专业开发经验，熟悉 React-based chat UIs、Dashboard 及 Monorepo 设置（如 Turborepo, Nx）。\n- **MCP 协议**: 具有 Model Context Protocol 的实操经验，熟悉 MCP Apps 规范，曾实现过 MCP 客户端和服务端，理解 MCP Gateway 架构。\n- **UI 集成**: 有在 React\u002FVue 中集成第三方 UI 组件的经验，熟悉组件库和设计系统，构建过可配置的聊天界面小部件系统。\n- **测试与演示**: 能够制作视频演示（使用 Loom\u002FOBS 等工具）作为提交成果的一部分。",[180,185,190,195,200,205,210,215,220,225,230,235,240,245,250,255,260,265,270,275],{"id":181,"version":182,"summary_zh":183,"released_at":184},222661,"platform-v1.2.10","## [1.2.10](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.9...platform-v1.2.10) (2026-04-11)\n\n\n### 功能特性\n\n* 使 PostHog 分析可配置（[#3707](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3707)）（[7ae9101](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F7ae9101aad7ce1c1cc5843f1c2d0e67fa7d5132f)）\n\n\n### 错误修复\n\n* `\u002Fllm\u002Fcosts` 表格滚动问题（[#3722](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3722)）（[6a42ba8](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6a42ba8de8ab971295e96c499d887c7d790a691d)）\n* 为网关 slug 应用 MCP OAuth 生命周期（[#3711](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3711)）（[362aaec](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F362aaec5126ce828727961ed46207e998c5f6627)）\n* Bedrock 工具名称编码问题（[#3706](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3706)）（[0e2c2d1](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F0e2c2d1521c3e0a86fa573e95f6c8695867dc6dd)）\n* 成本时间范围及表面限制重置设置（[#3709](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3709)）（[6e4154b](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6e4154b292bb4fc8c0abce36d7f6de7b425a5859)）\n* Jira OAuth 发现覆盖问题（[#3721](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3721)）（[2c4cf8f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F2c4cf8f39248272cc2b97b3752870c07914a6c2a)）\n* OIDC 发现中用于 IdP 注册的可信来源（[#3714](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3714)）（[adb5f5e](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fadb5f5edb39868ede3091ec61324b2872abb1385)）\n* 在分叉时保留共享聊天代理（[#3715](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3715)）（[252edfc](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F252edfc0178e60f975ba4597a0e7154a30312aaf)）\n* 重排序模型下拉菜单标签（[#3704](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3704)）（[ebd1c8a](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Febd1c8a1268d0a55a897628e57a427ffd21b8458)）\n* 会话日志加载状态（[#3712](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3712)）（[ffba126](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fffba126525dc0824ebe686817f97895219944bad)）\n\n\n### 其他杂项任务\n\n* **ci:** 添加 ID-JAG MCP 端到端测试（[#3702](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3702)）（[1a5078a](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F1a5078a7a65134be0ee0009b3710c79d256034ee)）\n* **deps:** 将 \u002Fplatform\u002Ffrontend 中的 Next.js 从 16.1.7 升级至 16.2.3（[#3708](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3708)）（[d47967c](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fd47967cf4635804c79951668813200831cb0af1a)）\n* 使用中性标记前缀并兼容旧版系统（[#3719](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3719)）（[db5929c](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fdb5929cb83d4aaef565836f8a934201b6396fbff)）","2026-04-11T04:55:52",{"id":186,"version":187,"summary_zh":188,"released_at":189},222662,"platform-v1.2.9","## [1.2.9](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.8...platform-v1.2.9) (2026-04-10)\n\n\n### 错误修复\n\n* 为检查器资源应用 MCP OAuth 的有效期（[#3701](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3701)）（[e4a4592](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe4a45927dcbeb220b58f758c4ebea28175b00a59)）","2026-04-10T16:34:40",{"id":191,"version":192,"summary_zh":193,"released_at":194},222663,"platform-v1.2.8","## [1.2.8](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.7...platform-v1.2.8) (2026-04-10)\n\n\n### 功能特性\n\n* 添加可配置的 MCP OAuth 令牌有效期设置 ([#3685](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3685)) ([e68db34](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe68db34053208ce672f4e3f01ae1f60f8c2f7122))\n\n\n### Bug 修复\n\n* 规范化跨消息的悬空工具调用 ([#3684](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3684)) ([36a731c](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F36a731c24df4893d09d5e65e3eb049647a239600))\n* 移除未使用的 vitest CI 标志位 ([#3694](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3694)) ([9b606f5](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9b606f5b33ce3be5bb4f322d03441acf4405a055))\n* 稳定端到端 CI 配置 ([#3693](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3693)) ([14bbeb5](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F14bbeb5b2deb2032b6baea55985ae108b09322d8))\n\n\n### 其他杂项工作\n\n* 将 API 端到端覆盖率移至路由测试，并稳定 CI ([#3683](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3683)) ([e4f70e3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe4f70e37b943085f97b7c9f4beb21a7b91e8d2b5))","2026-04-10T03:35:46",{"id":196,"version":197,"summary_zh":198,"released_at":199},222664,"platform-v1.2.7","## [1.2.7](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.6...platform-v1.2.7) (2026-04-08)\n\n\n### 杂项事务\n\n* 降低 Sentry 链路的噪声量 ([#3681](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3681)) ([650ae07](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F650ae075654c2790f64e854a73dd3574c3860ea9))","2026-04-08T20:01:26",{"id":201,"version":202,"summary_zh":203,"released_at":204},222665,"platform-v1.2.6","## [1.2.6](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.5...platform-v1.2.6) (2026-04-08)\n\n\n### 功能特性\n\n* **helm:** 改进 initContainers 配置选项 ([#3680](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3680)) ([8acd982](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F8acd982cb4c07de448c314ac52ef00aa32828b2b))\n\n\n### 问题修复\n\n* MCP 工具分配作用域 ([#3675](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3675)) ([f2687c3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Ff2687c31724297d3c9a2b0a9040859616576c120))\n* 重新生成 API 客户端以移除硬编码的 localhost baseUrl ([#3679](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3679)) ([7b01f39](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F7b01f397026d2a1873fec4cc427f1bf759998236))\n\n\n### 依赖更新\n\n* 将 \u002Fplatform 中的 drizzle-orm 从 0.45.1 升级至 0.45.2 ([#3668](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3668)) ([b27a61a](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fb27a61a6994f4f8c957020777b0400bc21327086))\n\n\n### 代码重构\n\n* 移除 A2A 和令牌选择路径中的硬编码 userIsAgentAdmin ([#3677](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3677)) ([c54d5b7](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fc54d5b7f95d4eda9298955b5b4ea1fa0c84542da))\n\n\n### 其他杂项\n\n* **依赖:** 将 \u002Fplatform\u002Fmcp_server_docker_image 中的 @hono\u002Fnode-server 从 1.19.10 升级至 1.19.13 ([#3673](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3673)) ([58850c1](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F58850c1f634ebf00b3033ad57dd4d5b7aee26706))\n* **依赖:** 将 \u002Fplatform\u002Fe2e-tests\u002Ftest-mcp-servers\u002Fmcp-server-jwks-keycloak 中的 @hono\u002Fnode-server 从 1.19.12 升级至 1.19.13 ([#3674](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3674)) ([5c5ffdb](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F5c5ffdbc5ca8c527d2c7ece0aaa48a1ec7e0c2e1))\n* **依赖:** 将 \u002Fplatform\u002Fmcp_server_docker_image 中的 hono 从 4.12.7 升级至 4.12.12 ([#3669](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3669)) ([1447d5f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F1447d5f0a7eb480060ad5efb22aa7050751d22e6))\n* **依赖:** 将 \u002Fplatform\u002Fe2e-tests\u002Ftest-mcp-servers\u002Fmcp-server-jwks-keycloak 中的 hono 从 4.12.9 升级至 4.12.12 ([#3672](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3672)) ([b805ad6](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fb805ad67b00f2513d3b462e2ee9aff0153a60cb0))\n* 加强动态凭据解析 ([#3663](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3663)) ([daf2c84](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fdaf2c84d1ce6c18c183f526bc0ba0f3f45c44ef7))","2026-04-08T18:41:46",{"id":206,"version":207,"summary_zh":208,"released_at":209},222666,"platform-v1.2.5","## [1.2.5](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.4...platform-v1.2.5) (2026-04-08)\n\n\n### 功能特性\n\n* 为 mcp-gw 添加请求头白名单 ([#3658](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3658)) ([459c53b](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F459c53b6a2b94744ca1404dbacb0b433783e7efb))\n* 添加 Microsoft SharePoint 知识连接器 ([#3656](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3656)) ([5b2174d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F5b2174dd30041aab717e13a54ada79d0d7fc2cd8))\n* 支持 Azure Responses API 流程 ([#3666](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3666)) ([e9e3982](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe9e39829fab6a417a40a61f2f1724f6b05385df8))\n\n\n### Bug 修复\n\n* 禁用默认优化规则 ([#3637](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3637)) ([3264c69](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F3264c692a3acae68f254491dab74ff6c3bc9ea6a))\n\n\n### 依赖更新\n\n* 将 \u002Fplatform 中的 lodash-es 从 4.17.23 升级至 4.18.1 ([#3638](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3638)) ([714daec](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F714daec443b9fe8f59a1868b79e7ca427126e988))","2026-04-08T02:13:56",{"id":211,"version":212,"summary_zh":213,"released_at":214},222671,"platform-v1.2.0","## [1.2.0](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.40...platform-v1.2.0) (2026-03-31)\r\n\r\n\r\n### Features\r\n\r\n* add enterprise-managed credentials for MCP auth and tool execution ([#3516](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3516)) ([2d5820f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F2d5820f196e4ded4a5288e86092e47aba6ffe2a5))\r\n* expand chat file modalities and llm provider key visibility ([#3574](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3574)) ([8f0b46c](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F8f0b46c3f9caffcc8f2bb70fbc17f6110c6f3c00))\r\n\r\n\r\n### Bug Fixes\r\n\r\n* improve Slack thread context for chatops bot ([#3565](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3565)) ([c3c538f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fc3c538f0130572795d4855c317666cd1a5e68aca))\r\n* mcp apps layout ([#3572](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3572)) ([6b277cb](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6b277cb652d0609aa2586ca4ca0671ec0037a1b8))\r\n* mcp-apps localhost mode firing in prod ([#3566](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3566)) ([9f20035](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9f20035574fcf83a45a95e7a9356ce1b9a2ed91a))\r\n\r\n\r\n### Dependencies\r\n\r\n* bump handlebars from 4.7.8 to 4.7.9 in \u002Fplatform ([#3585](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3585)) ([c69d30d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fc69d30dd56e837ee6bcb7b05879d17bf814b065a))\r\n\r\n\r\n### Miscellaneous Chores\r\n\r\n* delete e2e tests ([#3601](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3601)) ([6799e9d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6799e9d0d68b7b36b8ec0026b27b75e62f6b0cc1))\r\n* refine agent UI controls ([#3570](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3570)) ([0d9a8ba](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F0d9a8ba052701005d4066248a3b8ab8457c0a619))\r\n* **release:** bump version ([b0daa82](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fb0daa82e5bfdd64b5e1c49c322eab9d7aa33c7d7))\r\n* stabilize remaining e2e and auth follow-ups ([#3559](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3559)) ([e4ee425](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe4ee4252a2774f367eacd6469e29eee6812b6f9d))","2026-03-31T11:43:30",{"id":216,"version":217,"summary_zh":218,"released_at":219},222672,"platform-v1.1.40","## [1.1.40](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.39...platform-v1.1.40) (2026-03-26)\r\n\r\n\r\n### Features\r\n\r\n* add MCP Apps (SEP-1865) support ([#2898](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F2898)) ([9e46b21](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9e46b21fef2e840e84ac887e5cb9b77c840851f0))\r\n\r\n\r\n### Bug Fixes\r\n\r\n* add sandbox proxying on nextjs ([#3554](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3554)) ([7192dbd](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F7192dbd043f5fe198aaec54aa29d84ac79a8db0d))\r\n* improve Dockerfile ([#3560](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3560)) ([9919c53](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9919c538b0f143cc7a0f0afe33f0f0521df159b5))\r\n\r\n\r\n### Performance Improvements\r\n\r\n* reduce MCP gateway auth query churn ([#3557](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3557)) ([236e063](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F236e0631f4747926f2b12a9669c589d348996d23))\r\n\r\n\r\n### Dependencies\r\n\r\n* bump fastify from 5.8.2 to 5.8.3 in \u002Fplatform ([#3543](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3543)) ([f367afb](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Ff367afb2b41da0ac981aef65c01f16cd383cf0d6))\r\n* bump jsdom from 28.1.0 to 29.0.0 in \u002Fplatform ([#3542](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3542)) ([95851be](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F95851be2432cd4f21c840d95ad69edc5010d2f35))\r\n\r\n\r\n### Miscellaneous Chores\r\n\r\n* add PostHog user identification ([#3528](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3528)) ([624b5b3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F624b5b3e0ffef448f4b642a72cbd72e58ce53bce))\r\n* harden CI supply chain, action pinning, and e2e transport ([#3533](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3533)) ([151f3db](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F151f3dbae71942e30534ea0a08eac964902ab69a))\r\n* improve MCP gateway resilience and permission-gated team UX ([#3550](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3550)) ([fc26dce](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Ffc26dce2872d89468ab5e255345419fb2fd5572c))\r\n* use `better-auth` trustedProviders resolver for SSO account linking ([#3537](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3537)) ([9b6a95d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9b6a95dc74b99f1a26effec13a2673ee845106ba))","2026-03-30T20:13:54",{"id":221,"version":222,"summary_zh":223,"released_at":224},222673,"platform-v1.1.39","## [1.1.39](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.38...platform-v1.1.39) (2026-03-24)\n\n\n### Bug Fixes\n\n* use rwx diagnostics storage in staging and package worker startup ([#3523](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3523)) ([3460611](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F34606110bab406f3da18845c0eca99da0181948e))\n\n\n### Miscellaneous Chores\n\n* **deps:** reduce docker image CVEs ([#3525](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3525)) ([ad08212](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fad08212f061a58d56d1b28586ad990f5f6b577a2))","2026-03-24T20:57:11",{"id":226,"version":227,"summary_zh":228,"released_at":229},222674,"platform-v1.1.38","## [1.1.38](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.37...platform-v1.1.38) (2026-03-24)\n\n\n### Features\n\n* add identity provider option to disable RP-Initiated Logout ([#3519](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3519)) ([5b88da4](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F5b88da4eeb04fbb4287e4551cedb9d43ddc42d2e))\n\n\n### Miscellaneous Chores\n\n* improve Sentry capture, Node diagnostics, and trace sampling ([#3520](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3520)) ([d0908dc](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fd0908dc5d25a77bb1f1bffe6336dc644eef57423))\n* load public auth config from the backend ([#3522](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3522)) ([adebc19](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fadebc1939b0b851761b537c072c9421ac7a528f3))\n* **refactor:** frontend lib into chat, tools, and hooks directories ([#3515](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3515)) ([ca84169](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fca84169e7cf68e9a58cf1a6c8421a3e23acba86e))","2026-03-24T18:59:17",{"id":231,"version":232,"summary_zh":233,"released_at":234},222675,"platform-v1.1.37","## [1.1.37](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.36...platform-v1.1.37) (2026-03-23)\n\n\n### Bug Fixes\n\n* Endless retry loop on provider error bug ([#3507](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3507)) ([64cbfe4](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F64cbfe49b9666c3fe904f34cb768f49080cb08f7))\n* probe Vertex Gemini fallback models when list only returns live audio ([#3504](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3504)) ([313c9e7](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F313c9e790152f308737ea4e514ef8b5f9b0fb9ad))\n* simplify chat model sync and provider fetchers ([#3508](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3508)) ([02ffa24](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F02ffa24d928df1e0422f821073f890678fc7b017))\n* strip non-ISO-8859-1 chars from chat agent ID header ([#3500](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3500)) ([e2c474c](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe2c474cc6bfbdf76c7cd6952834c02456584ab35))\n\n\n### Dependencies\n\n* bump the platform-dependencies group in \u002Fplatform with 15 updates ([#3478](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3478)) ([ab8b2f3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fab8b2f3ccb1800188250ddadd6cb4004e8afce97))\n\n\n### Miscellaneous Chores\n\n* improve MCP gateway auth performance and fix e2e coverage ([#3503](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3503)) ([ad2acfd](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fad2acfdd15d70a481e82920e3b0cc43a720c8a0d))","2026-03-23T23:54:47",{"id":236,"version":237,"summary_zh":238,"released_at":239},222676,"platform-v1.1.36","## [1.1.36](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.35...platform-v1.1.36) (2026-03-22)\n\n\n### Features\n\n* white-label built-in MCP server branding ([#3496](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3496)) ([c11c170](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fc11c170be5a874a853205443c0a57d77a2799a46)), closes [#3475](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3475)\n\n\n### Bug Fixes\n\n* generate https oauth metadata behind reverse proxies ([#3400](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3400)) ([cabc557](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fcabc557f18c6db3f3481a4129a42460c7eb43875))","2026-03-23T00:37:28",{"id":241,"version":242,"summary_zh":243,"released_at":244},222677,"platform-v1.1.35","## [1.1.35](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.34...platform-v1.1.35) (2026-03-20)\n\n\n### Bug Fixes\n\n* use listInferenceProfile to discover models in aws ([#3482](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3482)) ([98ecc15](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F98ecc15c3d9f3e33ced9cc371794aff2d95f7ed7))","2026-03-20T17:20:31",{"id":246,"version":247,"summary_zh":248,"released_at":249},222678,"platform-v1.1.34","## [1.1.34](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.33...platform-v1.1.34) (2026-03-20)\n\n\n### Bug Fixes\n\n* bring back chat autoscroll ([#3480](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3480)) ([c22e87d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fc22e87d2f2ffbf261e6fc90b8508e1a6edce3563))\n* Prevent leak of chosen model and agent for same browser ([#3485](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3485)) ([ffa301f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fffa301f992e6133f1595eba552f829a8f49571c2))\n* tighten agent builder MCP assignment and chat tool state handling ([#3477](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3477)) ([4b15ac8](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F4b15ac807d6282cfd9ed708f83e20dbc227e510e))\n\n\n### Dependencies\n\n* bump @microsoft\u002Fmsgraph-sdk-users from 1.0.0-preview.77 to 1.0.0-preview.80 in \u002Fplatform ([#3479](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3479)) ([5179285](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F5179285a6ab83124f18a5f5537a122a57e13f595))\n\n\n### Miscellaneous Chores\n\n* more debug info in mini view ([#3467](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3467)) ([7931be7](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F7931be701131f6e7345b1f116dd1fa334dd35bcb))\n* revert prevent leak of chosen model and agent for same browser ([#3486](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3486)) ([87369bd](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F87369bdeebbfc1da734bd234bfbdde96a549d412))","2026-03-20T12:39:59",{"id":251,"version":252,"summary_zh":253,"released_at":254},222679,"platform-v1.1.33","## [1.1.33](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.32...platform-v1.1.33) (2026-03-19)\n\n\n### Bug Fixes\n\n* fix github mcp installation ([#3465](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3465)) ([785873e](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F785873e4412bc9111745b3c6142906c76c5b95ff))\n\n\n### Code Refactoring\n\n* standardize dialogs, settings blocks, and time selectors ([#3470](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3470)) ([cdddefe](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fcdddefe4594d58333f57be2be5d691419d8fc539)), closes [#3462](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3462) [#3464](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3464)\n\n\n### Miscellaneous Chores\n\n* clarify current model ui ([#3472](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3472)) ([2c8662f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F2c8662f5aab893cd7d8d8743d2a4c6d71302d39d)), closes [#3463](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3463)\n* Improve TOON compression docs and LLM settings help ([#3466](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3466)) ([174a6f5](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F174a6f569a0d272549aad5c3fe78e64b388426a1)), closes [#2766](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F2766)","2026-03-19T18:07:25",{"id":256,"version":257,"summary_zh":258,"released_at":259},222680,"platform-v1.1.32","## [1.1.32](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.1.31...platform-v1.1.32) (2026-03-19)\n\n\n### Features\n\n* **docs:** annotate OpenAPI operations with RBAC metadata ([#3447](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3447)) ([e7bd55b](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe7bd55bb2c5925e20ab2fe7dd04a1d1c85811e41))\n\n\n### Bug Fixes\n\n* avoid duplicate metrics registration in web pods ([#3457](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3457)) ([07d5a45](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F07d5a45a52f7958eb488a63b6b0232816e9e6148))\n* avoid duplicate metrics registration on web startup ([#3456](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3456)) ([9097272](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9097272cd0f5937a6f261209b9e847877f1e21d9))\n* harden settings team members and api key creation ([#3450](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3450)) ([3c7d3c3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F3c7d3c3b3be73f1633a0bdcc1330567abb6bcdb4))\n* normalize anthropic to bedroc format ([#3448](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3448)) ([82044dd](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F82044dd39f99d308b8c77d6f85f76502dcf03a35))\n* restore Gemini tool progress ([#3454](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3454)) ([3d650e0](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F3d650e0ddd725583d2bd2aa5183cfcf33ffff313))\n* restore knowledge base and worker metrics dashboards ([#3453](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3453)) ([645244f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F645244ffe2d910aca1235588bcabd589eb40ca29))\n* tighten knowledge base dashboard aggregations ([#3459](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3459)) ([abb1838](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fabb1838f111e0aba826353cad1ae9a743156d887))\n\n\n### Miscellaneous Chores\n\n* Refactor dual LLM into built-in agents ([#3455](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3455)) ([ac67158](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fac67158b672d1ca93f67e7d23249875752b9f568))\n* remove LLM proxy mock clients ([#3452](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3452)) ([e32bf9e](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe32bf9e478df5cede9f44076a38d436a8bead80d))","2026-03-19T13:56:37",{"id":261,"version":262,"summary_zh":263,"released_at":264},222667,"platform-v1.2.4","## [1.2.4](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.3...platform-v1.2.4) (2026-04-07)\n\n\n### 功能特性\n\n* 添加 Azure AI Foundry（Azure OpenAI）作为 LLM 提供者 ([#3659](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3659)) ([6a0b207](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6a0b207ae62a9e6e65d14111d5a92e77c32ed8ba))\n* 添加连接器级别的知识源 ACL ([#3416](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3416)) ([6039794](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F60397949df985816db87d75339d8a596ddeb6527))\n* 在 JWKS 认证中优先使用上游凭据而非 JWT 传播 ([#3061](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3061)) ([68ef87a](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F68ef87a624d057aba4c7a2cb54c6c6691a8d91cb))\n\n\n### 文档\n\n* 添加平台概览页面 ([#3625](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3625)) ([9b914e3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F9b914e389bb3567809fccab6fbe881c0d3ea990a))\n\n\n### 其他杂项工作\n\n* 添加 MCP 会话 TTL 清理，并修复应用指标的 Pod 聚合问题 ([#3647](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3647)) ([d1abfd9](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fd1abfd9194d1c6b023ab0e3f326e15aca5c62e63))\n* 修复 OAuth 同意及 MCP 认证相关说明 ([#3654](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3654)) ([5f4f210](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F5f4f210a9d310eb927c347909b9cbb45253b2ef5))\n* 将 `minimum-release-age` 值提高至 7 天 ([#3642](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3642)) ([655dcb2](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F655dcb21f1f7dc8bd85d435cb33071c1e76cf6bf))\n* MCP 网关 Slug ([#3652](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3652)) ([53f375f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F53f375f24da27594337dbf2d5903703ba636a247))","2026-04-07T02:35:18",{"id":266,"version":267,"summary_zh":268,"released_at":269},222668,"platform-v1.2.3","## [1.2.3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.2...platform-v1.2.3) (2026-04-03)\n\n\n### 功能特性\n\n* 添加 Notion 知识连接器 ([#3555](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3555)) ([808f5b7](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F808f5b7f7e7060a31c82b1929bc676b85f35619f))\n* 通过子代理委派传播工具护栏，并澄清范围访问控制文档 ([#3627](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3627)) ([1ad51ca](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F1ad51cab702acb1fbaebb8d5ca2b76f323ef8fc3))\n\n\n### 错误修复\n\n* 添加 MCP 认证扩展，并简化自托管认证流程 ([#3633](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3633)) ([95c3db9](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F95c3db99b028f8f517979a3ccdb90a8168ab804a))\n* 对齐知识库可观测性指标并推出默认配置 ([#3635](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3635)) ([de571d3](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fde571d30cd99ca52517a031012936e78dd78ef9f))\n* 格式化 playwright.config.ts 以通过 Biome CI 检查 ([#3631](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3631)) ([6f0df4d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F6f0df4d5d4721bb56131104eff8d7dbb046f7169))\n* 简化聊天状态及代理切换处理逻辑 ([#3636](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3636)) ([e2d70c1](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fe2d70c11a2af3e695f05f255fba0c22896d9564a))\n\n\n### 其他杂项工作\n\n* 更改 Playwright 工作进程 ([#3618](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3618)) ([bd4836d](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fbd4836d8b9fb5260a3baf0476109bb52df6dc5ed))","2026-04-03T13:45:52",{"id":271,"version":272,"summary_zh":273,"released_at":274},222669,"platform-v1.2.2","## [1.2.2](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.1...platform-v1.2.2) (2026-04-01)\n\n\n### 功能特性\n\n* 添加作用域内的聊天分享及规范化的聊天路由 ([#3616](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3616)) ([ba53dbf](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fba53dbfc8f6ec037f5ab0990ecf7134e16f41062))\n* 使知识库嵌入模型驱动化 ([#3611](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3611)) ([eb3d546](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Feb3d54661d45386f92880aec88d88af33714d279))\n\n\n### 错误修复\n\n* 修复聊天分享迁移顺序问题 ([#3622](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3622)) ([71f3e91](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F71f3e9147dafcaee05d18519162030bce89234ff))\n* 用户表限制问题 ([#3621](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3621)) ([18ce1d5](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F18ce1d5ef3933352de00996d3009156e66e4406f))\n\n\n### 其他杂项\n\n* 在护栏表格中正确渲染子代理工具 ([#3620](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3620)) ([53f1399](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F53f139912c475cd3492548521e75e20312f0190c))\n* 改进策略配置中的子代理 ([#3612](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3612)) ([d00b532](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fd00b5325d51be90dd6eb388d940cc14c3c913920))\n* 优化工具护栏、聊天流程以及知识\u002F模型的用户体验 ([#3624](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3624)) ([30d9e68](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F30d9e681cee6fac0255fb94e1590691cea006f3b))","2026-04-01T22:10:32",{"id":276,"version":277,"summary_zh":278,"released_at":279},222670,"platform-v1.2.1","## [1.2.1](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcompare\u002Fplatform-v1.2.0...platform-v1.2.1) (2026-03-31)\n\n\n### Bug修复\n\n* 添加SSE心跳机制，防止连接中断（[#3605](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3605)）（[8e20d6b](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F8e20d6b2c51eb76ffd44347ee2210e997ae94d0f)）\n* 在Gemini代理的透传路由中解析虚拟API密钥（[#3602](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3602)）（[d3538fc](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002Fd3538fcada767e78c351fa17162926fabe5ed1c0)）\n\n\n### 其他杂项工作\n\n* 当a2a需要人工审批时，阻止工具调用（[#3608](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3608)）（[3ae75b4](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F3ae75b448a41eb4f2741b4e38631f05bd6b786fb)）\n* **依赖更新：** 将\u002Fplatform\u002Fmcp_server_docker_image中的path-to-regexp从8.3.0升级到8.4.0（[#3587](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3587)）（[2965f9f](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F2965f9f50a55ffc216693e70d664f684d7f93673)）\n* 重构代理邮件触发器设置的用户界面体验（[#3581](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3581)）（[7343fd8](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F7343fd851037c33d39bc59a482465b377483d3c1)）\n* 在UI中使用“安全\u002F敏感”代替“可信\u002F不可信”上下文（[#3609](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fissues\u002F3609)）（[1a4c667](https:\u002F\u002Fgithub.com\u002Farchestra-ai\u002Farchestra\u002Fcommit\u002F1a4c6675ae8f8b5c4a90381ffdfa87f25264a775)）","2026-03-31T16:25:39"]