[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-amd--gaia":3,"tool-amd--gaia":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":79,"owner_twitter":79,"owner_website":80,"owner_url":81,"languages":82,"stars":123,"forks":124,"last_commit_at":125,"license":126,"difficulty_score":127,"env_os":128,"env_gpu":129,"env_ram":130,"env_deps":131,"category_tags":137,"github_topics":138,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":146,"updated_at":147,"faqs":148,"releases":181},2309,"amd\u002Fgaia","gaia","Build AI agents for your PC","GAIA 是 AMD 推出的开源 AI 智能体框架，专为在本地电脑构建和运行智能助手而设计。它最大的特点是完全离线运行，所有数据处理均在用户自己的 AMD Ryzen AI 硬件上完成，无需连接云端。这有效解决了敏感数据隐私泄露、高昂的 API 调用费用以及在无网络环境下无法部署 AI 等痛点，特别适合医疗、金融等对数据安全要求极高的场景。\n\n该工具主要面向开发者和技术研究人员，提供了灵活的 Python 和 C++ 双语言支持。通过简洁的代码接口，用户可以轻松定义智能体的行为逻辑、挂载自定义工具（如查询天气、检索文档），并快速集成语音交互、图像识别及文档问答（RAG）等高级功能。GAIA 深度优化了 AMD 处理器的 NPU 和核显性能，实现了高效的硬件加速推理。此外，其插件系统允许开发者将构建好的智能体打包发布，方便他人一键安装使用。如果你希望在保护隐私的前提下，低成本地在本地打造专属的 AI 应用，GAIA 是一个值得尝试的专业级开发框架。","# \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Famd_gaia_readme_9eee2f22e514.png\" alt=\"GAIA Logo\" width=\"64\" height=\"64\" style=\"vertical-align: middle;\"> GAIA: AI Agent Framework for AMD Ryzen AI\n\n[![GAIA CLI Tests](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Factions\u002Fworkflows\u002Ftest_gaia_cli.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Ftree\u002Fmain\u002Ftests \"Check out our cli tests\")\n[![Latest Release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Famd\u002Fgaia?include_prereleases)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases\u002Flatest \"Download the latest release\")\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Famd-gaia)](https:\u002F\u002Fpypi.org\u002Fproject\u002Famd-gaia\u002F)\n[![GitHub downloads](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fdownloads\u002Famd\u002Fgaia\u002Ftotal.svg)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases)\n[![OS - Windows](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOS-Windows-blue)](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart \"Windows installation\")\n[![OS - Linux](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOS-Linux-green)](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart \"Linux installation\")\n[![Python 3.10+](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPython-3.10+-blue?logo=python&logoColor=white)](https:\u002F\u002Fwww.python.org\u002F)\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20Community-7289DA?logo=discord&logoColor=white)](https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1392562559122407535\u002F1402013282495102997)\n\n**GAIA** is AMD's open-source framework for building intelligent AI agents that run **100% locally** on AMD Ryzen AI hardware. Keep your data private, eliminate cloud costs, and deploy in air-gapped environments—all with hardware-accelerated performance.\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart\">\u003Cstrong>Get Started →\u003C\u002Fstrong>\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## Why GAIA?\n\n| Feature | Description |\n|---------|-------------|\n| **100% Local** | All data stays on your machine—perfect for sensitive workloads and air-gapped deployments |\n| **Zero Cloud Costs** | No API fees, no usage limits, no subscriptions—unlimited AI at no extra cost |\n| **Privacy-First** | HIPAA-compliant, GDPR-friendly—ideal for healthcare, finance, and enterprise |\n| **Ryzen AI Optimized** | Hardware-accelerated inference using NPU + iGPU on AMD Ryzen AI processors |\n\n---\n\n## Build Your First Agent\n\n```python\nfrom gaia.agents.base.agent import Agent\nfrom gaia.agents.base.tools import tool\n\nclass MyAgent(Agent):\n    \"\"\"A simple agent with custom tools.\"\"\"\n\n    def _get_system_prompt(self) -> str:\n        return \"You are a helpful assistant.\"\n\n    def _register_tools(self):\n        @tool\n        def get_weather(city: str) -> dict:\n            \"\"\"Get weather for a city.\"\"\"\n            return {\"city\": city, \"temperature\": 72, \"conditions\": \"Sunny\"}\n\nagent = MyAgent()\nresult = agent.process_query(\"What's the weather in Austin?\")\nprint(result)\n```\n\n**[See the full quickstart guide →](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)**\n\n---\n\n## Key Capabilities\n\n- **Agent Framework** — Base class with tool orchestration, state management, and error recovery\n- **RAG System** — Document indexing and semantic search for Q&A\n- **Voice Integration** — Whisper ASR + Kokoro TTS for speech interaction\n- **Vision Models** — Extract text from images with Qwen3-VL-4B\n- **Plugin System** — Distribute agents via PyPI with auto-discovery\n- **Web UI Packaging** — Generate modern interfaces for your agents\n\n---\n\n## C++ Framework\n\nA C++17 port of the GAIA base agent framework is available under [`cpp\u002F`](cpp\u002FREADME.md). It implements the same agent loop, tool registry, and MCP client interface without any Python dependency — suitable for embedding in native applications or resource-constrained environments.\n\n```cpp\n#include \u003Cgaia\u002Fagent.h>\n\nclass MyAgent : public gaia::Agent {\nprotected:\n    std::string getSystemPrompt() const override {\n        return \"You are a helpful assistant.\";\n    }\n};\n```\n\n**[C++ build and usage instructions →](cpp\u002FREADME.md)**\n\n---\n\n## Quick Install\n\n```bash\npip install amd-gaia\n```\n\nFor complete setup instructions including Lemonade Server, see the **[Quickstart Guide](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)**.\n\n---\n\n## System Requirements\n\n| Requirement | Minimum | Recommended |\n|-------------|---------|-------------|\n| **Processor** | AMD Ryzen AI 300-series | AMD Ryzen AI Max+ 395 |\n| **OS** | Windows 11, Linux | - |\n| **RAM** | 16GB | 64GB |\n\n---\n\n## Documentation\n\n- **[Quickstart](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)** — Build your first agent in 10 minutes\n- **[SDK Reference](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fsdk)** — Complete API documentation\n- **[Guides](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fguides)** — Chat, Voice, RAG, and more\n- **[FAQ](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freference\u002Ffaq)** — Frequently asked questions\n\n---\n\n## Releases\n\nSee the full [Release Notes](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freleases) on the documentation site, or browse [GitHub Releases](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases).\n\n### Release Process\n\nTo publish a new release (e.g. `v0.17.0`), create a release PR that updates these 3 files:\n\n| # | File | What to change |\n|---|------|----------------|\n| 1 | `src\u002Fgaia\u002Fversion.py` | Set `__version__ = \"0.17.0\"` |\n| 2 | `docs\u002Freleases\u002Fv0.17.0.mdx` | Create release notes (see [format guide](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freleases)) |\n| 3 | `docs\u002Fdocs.json` | **(a)** Add `\"releases\u002Fv0.17.0\"` to the Releases tab pages array, **(b)** update the navbar label to `\"v0.17.0 · Lemonade X.Y.Z\"` |\n\nThen merge and tag:\n\n```bash\ngit tag v0.17.0 && git push origin v0.17.0\n```\n\nCI validates all three files are consistent with the tag before publishing to [GitHub Releases](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases) and [PyPI](https:\u002F\u002Fpypi.org\u002Fproject\u002Famd-gaia\u002F).\n\n---\n\n## Contributing\n\nWe welcome contributions! See our [Contributing Guide](CONTRIBUTING.md) for details.\n\n- **Build agents** in your own repository using GAIA as a dependency\n- **Improve the framework** — check [GitHub Issues](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues) for open tasks\n- **Add documentation** — examples, tutorials, and guides\n\n---\n\n## Contact\n\n- **Email**: [gaia@amd.com](mailto:gaia@amd.com)\n- **Discord**: [Join our community](https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1392562559122407535\u002F1402013282495102997)\n- **Issues**: [GitHub Issues](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues)\n\n---\n\n## License\n\n[MIT License](.\u002FLICENSE.md)\n\nCopyright(C) 2024-2025 Advanced Micro Devices, Inc. All rights reserved.\nSPDX-License-Identifier: MIT\n","# \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Famd_gaia_readme_9eee2f22e514.png\" alt=\"GAIA Logo\" width=\"64\" height=\"64\" style=\"vertical-align: middle;\"> GAIA：面向AMD Ryzen AI的AI智能体框架\n\n[![GAIA CLI 测试](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Factions\u002Fworkflows\u002Ftest_gaia_cli.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Ftree\u002Fmain\u002Ftests \"查看我们的 CLI 测试\")\n[![最新版本](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Famd\u002Fgaia?include_prereleases)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases\u002Flatest \"下载最新版本\")\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Famd-gaia)](https:\u002F\u002Fpypi.org\u002Fproject\u002Famd-gaia\u002F)\n[![GitHub 下载量](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fdownloads\u002Famd\u002Fgaia\u002Ftotal.svg)](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases)\n[![操作系统 - Windows](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOS-Windows-blue)](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart \"Windows 安装指南\")\n[![操作系统 - Linux](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOS-Linux-green)](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart \"Linux 安装指南\")\n[![Python 3.10+](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPython-3.10+-blue?logo=python&logoColor=white)](https:\u002F\u002Fwww.python.org\u002F)\n[![许可证：MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20Community-7289DA?logo=discord&logoColor=white)](https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1392562559122407535\u002F1402013282495102997)\n\n**GAIA** 是 AMD 开源的框架，用于构建在 AMD Ryzen AI 硬件上 **100% 本地运行** 的智能 AI 智能体。保护您的数据隐私，无需支付云端费用，并可在气隙环境中部署——所有这些都得益于硬件加速性能。\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart\">\u003Cstrong>开始使用 →\u003C\u002Fstrong>\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## 为什么选择 GAIA？\n\n| 特性 | 描述 |\n|---------|-------------|\n| **100% 本地** | 所有数据均保留在您的设备上——非常适合敏感工作负载和气隙部署 |\n| **零云端成本** | 无 API 费用、无使用限制、无订阅——无限量 AI 使用无需额外费用 |\n| **隐私优先** | 符合 HIPAA 和 GDPR 标准——非常适合医疗、金融和企业级应用 |\n| **Ryzen AI 优化** | 利用 AMD Ryzen AI 处理器上的 NPU + iGPU 进行硬件加速推理 |\n\n---\n\n## 构建您的第一个智能体\n\n```python\nfrom gaia.agents.base.agent import Agent\nfrom gaia.agents.base.tools import tool\n\nclass MyAgent(Agent):\n    \"\"\"一个带有自定义工具的简单智能体。\"\"\"\n\n    def _get_system_prompt(self) -> str:\n        return \"您是一位乐于助人的助手。\"\n\n    def _register_tools(self):\n        @tool\n        def get_weather(city: str) -> dict:\n            \"\"\"获取某个城市的天气信息。\"\"\"\n            return {\"city\": city, \"temperature\": 72, \"conditions\": \"晴朗\"}\n\nagent = MyAgent()\nresult = agent.process_query(\"奥斯汀的天气如何？\")\nprint(result)\n```\n\n**[查看完整快速入门指南 →](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)**\n\n---\n\n## 核心功能\n\n- **智能体框架** — 基础类，具备工具编排、状态管理和错误恢复功能\n- **RAG 系统** — 文档索引与语义搜索，用于问答任务\n- **语音集成** — Whisper ASR + Kokoro TTS，实现语音交互\n- **视觉模型** — 使用 Qwen3-VL-4B 从图像中提取文本\n- **插件系统** — 通过 PyPI 分发智能体，支持自动发现\n- **Web UI 打包** — 为您的智能体生成现代化界面\n\n---\n\n## C++ 框架\n\nGAIA 基础智能体框架的 C++17 版本已在 [`cpp\u002F`](cpp\u002FREADME.md) 中提供。它实现了相同的智能体循环、工具注册表和 MCP 客户端接口，且无需任何 Python 依赖——非常适合嵌入原生应用程序或资源受限环境。\n\n```cpp\n#include \u003Cgaia\u002Fagent.h>\n\nclass MyAgent : public gaia::Agent {\nprotected:\n    std::string getSystemPrompt() const override {\n        return \"您是一位乐于助人的助手。\";\n    }\n};\n```\n\n**[C++ 构建与使用说明 →](cpp\u002FREADME.md)**\n\n---\n\n## 快速安装\n\n```bash\npip install amd-gaia\n```\n\n有关包括 Lemonade Server 在内的完整设置说明，请参阅 **[快速入门指南](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)**。\n\n---\n\n## 系统要求\n\n| 要求 | 最低配置 | 推荐配置 |\n|-------------|---------|-------------|\n| **处理器** | AMD Ryzen AI 300 系列 | AMD Ryzen AI Max+ 395 |\n| **操作系统** | Windows 11、Linux | - |\n| **内存** | 16GB | 64GB |\n\n---\n\n## 文档\n\n- **[快速入门](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)** — 10 分钟内构建您的第一个智能体\n- **[SDK 参考](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fsdk)** — 完整的 API 文档\n- **[指南](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fguides)** — 聊天、语音、RAG 等\n- **[常见问题解答](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freference\u002Ffaq)** — 常见问题解答\n\n---\n\n## 发布记录\n\n请访问文档网站查看完整的 [发布说明](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freleases)，或浏览 [GitHub 发布页面](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases)。\n\n### 发布流程\n\n要发布新版本（例如 `v0.17.0`），请创建一个包含以下 3 个文件更新的发布 PR：\n\n| # | 文件 | 修改内容 |\n|---|------|----------------|\n| 1 | `src\u002Fgaia\u002Fversion.py` | 设置 `__version__ = \"0.17.0\"` |\n| 2 | `docs\u002Freleases\u002Fv0.17.0.mdx` | 创建发布说明（请参考 [格式指南](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Freleases)） |\n| 3 | `docs\u002Fdocs.json` | **(a)** 将 `\"releases\u002Fv0.17.0\"` 添加到 Releases 标签页数组中，**(b)** 更新导航栏标签为 `\"v0.17.0 · Lemonade X.Y.Z\"` |\n\n然后合并并打标签：\n\n```bash\ngit tag v0.17.0 && git push origin v0.17.0\n```\n\nCI 会验证这三个文件是否与标签一致，随后才会发布到 [GitHub Releases](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Freleases) 和 [PyPI](https:\u002F\u002Fpypi.org\u002Fproject\u002Famd-gaia\u002F)。\n\n---\n\n## 贡献\n\n我们欢迎各位贡献！详情请参阅我们的 [贡献指南](CONTRIBUTING.md)。\n\n- 使用 GAIA 作为依赖项，在您自己的仓库中构建智能体\n- 改进框架——查看 [GitHub Issues](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues) 以了解当前待办事项\n- 补充文档——示例、教程和指南\n\n---\n\n## 联系方式\n\n- **邮箱**: [gaia@amd.com](mailto:gaia@amd.com)\n- **Discord**: [加入我们的社区](https:\u002F\u002Fdiscord.com\u002Fchannels\u002F1392562559122407535\u002F1402013282495102997)\n- **问题反馈**: [GitHub Issues](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues)\n\n---\n\n## 许可证\n\n[MIT 许可证](.\u002FLICENSE.md)\n\n版权所有 © 2024–2025 Advanced Micro Devices, Inc. 保留所有权利。\nSPDX 许可标识符：MIT","# GAIA 快速上手指南\n\nGAIA 是 AMD 推出的开源 AI 智能体框架，专为 **AMD Ryzen AI** 硬件优化。它支持在本地运行 100% 私有的 AI 应用，利用 NPU 和 iGPU 进行硬件加速推理，无需云端依赖，适合对数据隐私要求极高的场景。\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### 系统要求\n| 组件 | 最低要求 | 推荐配置 |\n| :--- | :--- | :--- |\n| **处理器** | AMD Ryzen AI 300 系列 | AMD Ryzen AI Max+ 395 |\n| **操作系统** | Windows 11 或 Linux | - |\n| **内存 (RAM)** | 16 GB | 64 GB |\n| **Python 版本** | Python 3.10 或更高 | - |\n\n### 前置依赖\n- 确保已安装支持 AMD Ryzen AI 的驱动程序及运行时环境（通常随官方驱动安装包提供）。\n- 建议配置好 Python 虚拟环境以避免依赖冲突。\n\n## 安装步骤\n\n使用 pip 直接安装 GAIA 核心库：\n\n```bash\npip install amd-gaia\n```\n\n> **提示**：完整的部署可能还需要配置 Lemonade Server 以支持模型服务。详细的环境搭建（包括国内网络优化建议）请参考官方 [快速入门指南](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fquickstart)。\n\n## 基本使用\n\n以下是一个最简单的示例，展示如何定义一个带有自定义工具的智能体并处理查询：\n\n```python\nfrom gaia.agents.base.agent import Agent\nfrom gaia.agents.base.tools import tool\n\nclass MyAgent(Agent):\n    \"\"\"一个带有自定义工具的简单智能体。\"\"\"\n\n    def _get_system_prompt(self) -> str:\n        return \"You are a helpful assistant.\"\n\n    def _register_tools(self):\n        @tool\n        def get_weather(city: str) -> dict:\n            \"\"\"获取城市天气。\"\"\"\n            return {\"city\": city, \"temperature\": 72, \"conditions\": \"Sunny\"}\n\n# 实例化智能体\nagent = MyAgent()\n\n# 处理查询\nresult = agent.process_query(\"What's the weather in Austin?\")\nprint(result)\n```\n\n运行上述代码后，GAIA 将自动调用注册的 `get_weather` 工具来回答关于天气的问题，所有推理过程均在本地 AMD Ryzen AI 硬件上完成。\n\n更多高级功能（如 RAG 检索增强、语音交互、视觉模型集成等）请参阅官方 [SDK 文档](https:\u002F\u002Famd-gaia.ai\u002Fdocs\u002Fsdk)。","某金融合规团队需要在完全隔离的内网环境中，利用本地文档快速构建一个能回答内部政策问答并支持语音交互的智能助手。\n\n### 没有 gaia 时\n- **数据泄露风险高**：为满足隐私合规（如 GDPR\u002FHIPAA），必须禁止使用云端大模型，导致无法利用先进的 AI 能力处理敏感金融数据。\n- **硬件算力闲置**：团队虽配备了搭载 NPU 的 AMD Ryzen AI 笔记本，但缺乏专用框架调用本地异构算力，只能依赖低效的 CPU 推理。\n- **开发集成繁琐**：从零搭建包含文档检索（RAG）、语音识别（Whisper）和语音合成（TTS）的全链路系统，需整合多个独立库，维护成本极高。\n- **部署门槛高**：在无外网的“气隙”环境中安装依赖复杂，且难以将智能体打包成带界面的应用分发给业务人员。\n\n### 使用 gaia 后\n- **100% 本地隐私安全**：gaia 确保所有数据仅在本地 AMD 硬件上流转，无需联网即可满足最严苛的金融合规要求，彻底消除数据出域风险。\n- **NPU 硬件加速**：自动调用 Ryzen AI 的 NPU 和 iGPU 进行推理，显著提升响应速度，同时降低功耗，让老旧设备也能流畅运行大模型。\n- **全功能开箱即用**：内置 RAG 文档索引、Whisper 语音识别及 Kokoro 语音合成模块，开发者仅需几行代码即可组装具备听、说、查能力的智能体。\n- **一键打包分发**：利用 gaia 的 Web UI 封装功能，可直接生成带有现代界面的独立应用，方便在内网环境中分发给非技术背景的合规专员使用。\n\ngaia 让企业在零云成本和高隐私标准下，轻松将本地 AMD 硬件转化为强大的私有化 AI 生产力中心。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Famd_gaia_9eee2f22.png","amd","AMD","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Famd_f16a8927.png","",null,"http:\u002F\u002Fwww.amd.com","https:\u002F\u002Fgithub.com\u002Famd",[83,87,91,95,99,103,107,111,115,119],{"name":84,"color":85,"percentage":86},"Python","#3572A5",66.8,{"name":88,"color":89,"percentage":90},"JavaScript","#f1e05a",8.1,{"name":92,"color":93,"percentage":94},"C++","#f34b7d",7.1,{"name":96,"color":97,"percentage":98},"TypeScript","#3178c6",6.5,{"name":100,"color":101,"percentage":102},"HTML","#e34c26",5.1,{"name":104,"color":105,"percentage":106},"CSS","#663399",4.1,{"name":108,"color":109,"percentage":110},"PowerShell","#012456",1.1,{"name":112,"color":113,"percentage":114},"Jupyter Notebook","#DA5B0B",0.5,{"name":116,"color":117,"percentage":118},"Shell","#89e051",0.4,{"name":120,"color":121,"percentage":122},"CMake","#DA3434",0.2,1044,74,"2026-04-03T08:05:48","MIT",4,"Windows, Linux","必需 AMD Ryzen AI 处理器（集成 NPU + iGPU），推荐 AMD Ryzen AI Max+ 395，不支持 NVIDIA CUDA","最低 16GB，推荐 64GB",{"notes":132,"python":133,"dependencies":134},"该框架专为 AMD Ryzen AI 硬件设计，支持 100% 本地运行以保护隐私；包含 C++17 版本可供无 Python 依赖环境使用；需配合 Lemonade Server 进行完整设置。","3.10+",[135,136],"amd-gaia","Lemonade Server",[14,13,15],[139,75,140,141,142,143,144,145],"ai","genai","ryzenai","agents","aipc","local","privacy","2026-03-27T02:49:30.150509","2026-04-06T05:27:13.168290",[149,154,158,162,167,172,177],{"id":150,"question_zh":151,"answer_zh":152,"source_url":153},10579,"GAIA 的混合模式（Hybrid Mode）是如何工作的？NPU 和 GPU 分别负责什么任务？","在混合模式下，提示词处理（Prompt Processing\u002F输入阶段）由 NPU 完成，而令牌生成（Token Generation\u002F输出阶段）由 GPU 负责。如果您输入的提示词较短但生成的令牌很多，GPU 的利用率会远高于 NPU；如果您提供较长的提示词（1000+ tokens），则会看到更明显的 NPU 激活。NPU 快速完成工作是为了让 GPU 能尽快利用 KV Cache 进行生成。","https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues\u002F51",{"id":155,"question_zh":156,"answer_zh":157,"source_url":153},10580,"为什么我的 NPU 利用率很低或者几乎没有被使用？","这通常是因为输入的提示词（Prompt）太短。NPU 主要用于处理输入阶段，如果提示词很短，NPU 会在 1-2 秒内迅速完成工作（利用率瞬间飙升后回落），随后大部分时间由 GPU 进行生成任务。建议尝试使用超过 1000 个 token 的长提示词，这样可以观察到更持续和明显的 NPU 利用率。",{"id":159,"question_zh":160,"answer_zh":161,"source_url":153},10581,"Ryzen 8000 系列笔记本电脑支持混合模式（NPU 加速 LLM）吗？","不支持。Ryzen 8000 系列笔记本电脑中的 NPU 算力较小，无法有效运行大型语言模型（LLM），因此 Ryzen AI LLM 软件不支持该系列。建议在这些设备上使用 GGUF 格式的模型，通过 llamacpp + Vulkan 后端在 GPU 上运行以获得良好的性能。",{"id":163,"question_zh":164,"answer_zh":165,"source_url":166},10582,"安装 GAIA 后提示成功，但 Lemonade Server 无法启动或找不到组件怎么办？","这是一个已知问题，通常在旧版本中出现。请尝试升级到最新版本的 GAIA（例如 v0.8.4 或更高版本），该问题在后续发布中已修复。如果更新后问题依旧，请重新打开 Issue 并提供详细日志。","https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues\u002F41",{"id":168,"question_zh":169,"answer_zh":170,"source_url":171},10583,"AMD Ryzen 7 8840U\u002F8845HS 处理器支持 GAIA 的混合模式吗？","目前官方软件尚未直接支持 8840U\u002F8845HS 的 NPU 进行 LLM 加速。虽然这些芯片内置 NPU，但当前的 GAIA 混合设置主要针对更新的架构。对于 8845HS 等设备，建议使用 Ollama 或 LM Studio 等工具通过 GPU 实现加速，或者关注 AMD IRON 项目以了解底层的 NPU 开发原语。","https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues\u002F20",{"id":173,"question_zh":174,"answer_zh":175,"source_url":176},10584,"在 Linux 上运行 gaiaui 时出现白屏或 'TypeError: Parameter.make_metavar() missing...' 错误如何解决？","该错误通常与 GAIA UI (RAUX backend) 的特定版本有关。维护者指出 GAIA UI 是在 https:\u002F\u002Fgithub.com\u002Faigdat\u002Fraux 维护并集成到 GAIA 中的。如果遇到此问题，请确保您使用的是最新版本的 GAIA 安装包。如果问题是通过 Git Clone 安装的，请尝试拉取最新代码或提交 PR 修复。该问题在较新的版本中已被标记为 resolved。","https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fissues\u002F84",{"id":178,"question_zh":179,"answer_zh":180,"source_url":153},10585,"目前支持的混合模型最大参数量是多少？未来会支持更大的模型吗？","目前在 Gaia 模型服务器（Lemonade Server）上可用的最大混合模型约为 8B 参数。支持更大参数的通用模型正在进行中。在此期间，您可以加载任何喜欢的 GGUF 格式模型，它们将在 GPU 上运行（已有演示支持如 Llama-3-Scout 等超过 100B 参数的模型）。此外，部分模型（如 Qwen1.5-7B）的提示词限制已放宽至 3000 tokens。",[182,187,192,197,202,207,212,217,222,227,232,237,242,247,252,257,262,267,272,277],{"id":183,"version":184,"summary_zh":185,"released_at":186},71129,"v0.17.1","# GAIA v0.17.1 Release Notes\r\n\r\nGAIA v0.17.1 hardens the framework with better version safety, richer eval insights, and a smoother Agent UI setup experience. The C++ agent framework also gains SSE streaming, runtime reconfiguration, and CI benchmarks.\r\n\r\n**Why upgrade:**\r\n- **Catch Lemonade mismatches early** — GAIA now warns you at startup if your running Lemonade Server version differs from the expected version, so you fix compatibility issues before they bite\r\n- **See where eval time goes** — Per-turn performance data rolls up into scenario-level summaries in the eval scorecard, making it easy to spot slow steps\r\n- **Agent UI works out of the box** — `gaia init` now builds the frontend automatically, so you no longer need a manual `npm run build` step\r\n\r\n---\r\n\r\n## What's New\r\n\r\n### Lemonade Version Mismatch Warning\r\n\r\nGAIA now detects when your running Lemonade Server version differs from the expected version and displays a clear warning with upgrade instructions (PR #637). No more silent incompatibilities.\r\n\r\n### Eval Performance Tracking\r\n\r\nThe eval scorecard now aggregates per-turn performance data — token counts, latency, throughput — into scenario-level summaries, so you can identify bottlenecks across multi-step evaluations (PR #637).\r\n\r\n### MCP Inference Stats\r\n\r\nMCP responses now capture inference statistics from SSE events, giving you visibility into model performance during tool-augmented conversations (PR #637).\r\n\r\n### Agent UI Frontend Auto-Build\r\n\r\n`gaia init` now builds the Agent UI frontend automatically via a new `_ensure_webui_built()` helper extracted into `gaia\u002Fui\u002Fbuild.py` (PR #657). The Agent UI documentation prerequisites have also been corrected. Includes 7 new unit tests.\r\n\r\n---\r\n\r\n## C++ Agent Framework\r\n\r\nThe C++ agent framework continues to mature with three updates:\r\n\r\n- **SSE streaming** — `SseParser` for chunked SSE delivery, `LemonadeClient` streaming via httplib, and console streaming helpers. Enable with `GAIA_STREAMING=1`. Includes 18 new SSE parser tests (PR #518)\r\n- **Runtime reconfiguration** — `AgentConfig` serialization, environment variable overrides (`GAIA_BASE_URL`, `GAIA_MODEL_ID`, etc.), dynamic setters (`setModel()`, `setConfig()`, `setMaxSteps()`), and thread-safe tool enable\u002Fdisable via `ToolRegistry` (PR #531)\r\n- **Performance benchmarks** — New `benchmark_cpp.yml` CI workflow tracks binary size, startup time, loop latency, and memory. PR runs compare against cached baselines with configurable thresholds (PR #519)\r\n\r\n---\r\n\r\n## Bug Fixes\r\n\r\n- **MCP test isolation** — Added `isolate_home` autouse fixture to prevent unit tests from reading or writing to the real `~\u002F.gaia\u002Fmcp_servers.json` (PR #658)\r\n- **npm publish** — Switched to OIDC trusted publishing and removed `registry-url` to enable it (PRs #638, #639)\r\n- **Merge queue notifications** — Resolved phantom failures in the `merge-queue-notify` workflow (PR #640)\r\n\r\n---\r\n\r\n## Full Changelog\r\n\r\n**9 commits** since v0.17.0:\r\n\r\n- `780a711` - feat: Lemonade version mismatch warning, eval perf tracking, MCP stats (#637)\r\n- `7ed2db3` - feat(cpp): SSE streaming response support for C++ agent framework (#518)\r\n- `9c4101d` - feat(cpp): performance benchmarks and binary size tracking (#519)\r\n- `878a976` - feat(cpp): runtime configuration and dynamic reconfiguration (#531)\r\n- `e0e5695` - fix: isolate MCP unit tests from real ~\u002F.gaia\u002Fmcp_servers.json (#658)\r\n- `bb010a0` - fix: build Agent UI frontend in gaia init and fix doc prerequisites (#657)\r\n- `334b011` - fix: remove registry-url to enable OIDC trusted publishing (#639)\r\n- `776dc34` - fix: resolve merge-queue-notify phantom failures (#640)\r\n- `83a4db1` - fix: switch npm publish to OIDC trusted publishing (#638)\r\n\r\nFull Changelog: [v0.17.0...v0.17.1](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.17.0...v0.17.1)\r\n","2026-04-01T18:29:21",{"id":188,"version":189,"summary_zh":190,"released_at":191},71130,"v0.17.0","# GAIA v0.17.0 Release Notes\r\n\r\nRun AI agents locally on your PC — analyze documents, execute tools, and accomplish tasks without sending data to the cloud. GAIA v0.17.0 introduces the **Agent UI**, a privacy-first web application that puts a local AI agent on your AMD hardware.\r\n\r\n```bash\r\nnpm install -g @amd-gaia\u002Fagent-ui\r\ngaia-ui\r\n```\r\n\r\n**Why upgrade:**\r\n- **Keep sensitive documents on your machine** — Ask questions about medical records, contracts, financial data, or any of 53+ file types and get answers with page citations. Nothing leaves your PC — the agent runs 100% locally on your AMD hardware\r\n- **An AI agent you can trust** — Tool guardrails require your approval before the agent executes commands or writes files, so you stay in control of what happens on your machine\r\n- **Works on modest hardware** — A 78% smaller system prompt means GAIA now runs reliably on smaller models like Qwen3.5 without timeouts, so you don't need top-tier hardware to get started\r\n- **Access from anywhere** — Built-in ngrok tunnel lets you use your local GAIA instance from your phone or tablet while your data stays on your PC\r\n\r\nGet started with the [Agent UI guide](\u002Fguides\u002Fagent-ui) — install, launch, and run your first task in under 60 seconds.\r\n\r\n---\r\n\r\n## What's New\r\n\r\n### GAIA Agent UI\r\n\r\nA privacy-first web application for running AI agents locally on your AMD hardware (PR #428). Analyze documents, generate code, search files, execute tools, and accomplish tasks — without sending anything to the cloud.\r\n\r\n**What you can do:**\r\n- **Analyze your documents** — Drag-and-drop PDFs, Word docs, or any of 53+ file formats and get answers with page-level citations, powered by local RAG\r\n- **Execute tools safely** — The agent can run shell commands, write files, and use MCP tools — but only after you approve each action\r\n- **Search and browse files** — The agent can find files, explore directories, and locate content across your projects\r\n- **Access from your phone** — Built-in ngrok tunnel lets you use your local GAIA instance from any device\r\n- **Watch the agent think** — Real-time streaming with `\u003Cthink>` block rendering shows the agent's reasoning process inline\r\n- **Pick up where you left off** — Create, switch, and persist sessions with full history\r\n- **Monitor performance** — Hover tooltips show token counts, latency, and throughput metrics per response\r\n\r\n**Under the hood:**\r\n- FastAPI backend + React\u002FTypeScript frontend + Electron shell with SSE streaming\r\n- Redesigned Settings modal with system dashboard, model load\u002Fdownload actions, and live MCP server connection status with tool counts\r\n- Terminal-inspired design: typewriter welcome animation, pixelated AMD cursor with red glow, glassmorphism, smooth crossfade transitions (PR #568)\r\n- `prefers-reduced-motion` support — all animations respect OS accessibility settings\r\n- Path traversal prevention, SQL parameterization, and input validation throughout\r\n- 13+ bug fixes across backend, frontend, and integration layers\r\n\r\n```bash\r\nnpm install -g @amd-gaia\u002Fagent-ui\r\ngaia-ui\r\n```\r\n\r\nSee the [Agent UI guide](\u002Fguides\u002Fagent-ui) for full setup, prerequisites, and onboarding.\r\n\r\n---\r\n\r\n### Tool Execution Guardrails\r\n\r\nAI agents are powerful but can be unpredictable. This release adds a safety layer so you approve every sensitive action before it happens (PR #565):\r\n\r\n- **Confirmation popup** — **Allow**, **Deny**, or **Always Allow** before `run_shell_command` and other write\u002Fexecute tools\r\n- **60-second timeout** — Auto-denies if you don't respond within a minute\r\n- **Expanded coverage** — Extended to cover all write\u002Fexecute tools, not just shell commands (PR #604)\r\n\r\n---\r\n\r\n### Device Support Detection\r\n\r\nNot sure if your hardware is supported? GAIA now tells you upfront and offers workarounds (PR #593):\r\n\r\n- **Supported devices** — AMD Ryzen AI Max processors and AMD Radeon GPUs with ≥24 GB VRAM\r\n- **Clear banner messaging** — Shows your processor name and links to a GitHub feature-request\r\n- **`--base-url` flag** — Point to a remote Lemonade Server to use GAIA on any machine\r\n- **`GAIA_SKIP_DEVICE_CHECK=1`** — Environment variable override for advanced users\r\n\r\n---\r\n\r\n### System Prompt Optimization\r\n\r\nIf you previously experienced timeouts or slow first responses on smaller models, this release fixes that (PR #617):\r\n\r\n- **17,600 → 3,853 tokens (78% reduction)** — Two-tier RAG gating only injects document context when relevant, meaning 4–5× faster prompt processing\r\n- **Qwen3.5 timeouts eliminated** — Smaller prompt fits within context window of constrained models\r\n- **Timeout increased to 600s** — Prevents premature timeouts on complex queries\r\n\r\n---\r\n\r\n## Security\r\n\r\n- **Document upload vulnerability fixed** — Closed a TOCTOU race condition that could allow file substitution during upload. Now uses atomic `O_NOFOLLOW` + `fstat` validation and serializes concurrent uploads via per-file `asyncio.Lock` (PR #564)\r\n\r\n---\r\n\r\n## Bug Fixes\r\n\r\n- **LRU eviction silent failure**","2026-03-27T22:17:55",{"id":193,"version":194,"summary_zh":195,"released_at":196},71131,"v0.16.1","## What's Changed\n* Fixed broken links to amd-gaia.ai in README.md by @evanjrowley in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F497\n* Update Lemonade Server to v10.0.0 by @kovtcharov in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F498\n* Fix GAIA CLI Linux test hanging in merge queue by @kovtcharov in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F445\n* C++ framework: tool security — policies, confirmation, allowlisting, arg validation by @itomek in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F496\n* ci: add doc link checker, scope workflow triggers, fix broken links by @kovtcharov in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F499\n* Fix gaia init version checks and MSI reliability by @kovtcharov in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F478\n* Fix MCP config stacking and tool display naming by @itomek in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F477\n* feat(cpp): Process Analyst Agent with LemonadeClient and background monitoring by @itomek in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F426\n* Release v0.16.1 by @kovtcharov in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F516\n\n## New Contributors\n* @evanjrowley made their first contribution in https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F497\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.16.0...v0.16.1","2026-03-13T19:06:27",{"id":198,"version":199,"summary_zh":200,"released_at":201},71132,"v0.16.0","# GAIA v0.16.0 Release Notes\r\n\r\n**Major release** introducing the **C++17 Agent Framework** — a native port of the GAIA agent system that lets you build AI PC agents in pure C++ with no Python runtime required. Also includes security hardening across the Python codebase and a comprehensive documentation site reorganization.\r\n\r\n**TL;DR:**\r\n- **New: C++17 Agent Framework** — Full native agent framework with MCP support, tool registry, and state machine — no Python needed\r\n- **New: Health Agent & WiFi Agent** — Two production-ready C++ example agents for Windows system diagnostics\r\n- **New: CleanConsole** — Polished terminal UI for C++ agents with ANSI colors and word-wrap\r\n- **New: C++ CI\u002FCD Pipeline** — Cross-platform build + STX hardware integration tests\r\n- **Security: 37 alerts resolved** — 15 Dependabot + 22 code scanning alerts fixed across Python and JavaScript\r\n- **Docs: Site reorganization** — C++ Framework gets dedicated top-level section, SDK accuracy fixes, sidebar icons, 17+ broken links repaired\r\n\r\n---\r\n\r\n## What's New\r\n\r\n### C++17 Agent Framework\r\n\r\nA complete native port of the GAIA base agent framework under `cpp\u002F`. Build AI agents in pure C++17 with the same agent loop semantics as the Python `src\u002Fgaia\u002Fagents\u002Fbase\u002F` — no Python runtime required.\r\n\r\n**Key capabilities:**\r\n- **Agent state machine** — `PLANNING -> EXECUTING_PLAN -> DIRECT_EXECUTION -> ERROR_RECOVERY -> COMPLETION`\r\n- **ToolRegistry** with tool name resolution (unprefixed MCP name suffix match, case-insensitive fallback)\r\n- **MCP client** with cross-platform stdio transport (Win32 `CreateProcess` + POSIX `fork`\u002F`exec`), JSON-RPC 2.0, `mcp_\u003Cserver>_\u003Ctool>` naming convention\r\n- **JSON parsing** with 4-strategy fallback: direct parse, code-block extraction, bracket-matching, regex field extraction\r\n- **Parameter substitution** — `$PREV.field` \u002F `$STEP_N.field` across plan steps\r\n- **OpenAI-compatible HTTP client** via `cpp-httplib`\r\n\r\n```cpp\r\n#include \u003Cgaia\u002Fagent.h>\r\n\r\nint main() {\r\n    gaia::AgentConfig config;\r\n    config.baseUrl = \"http:\u002F\u002Flocalhost:8000\u002Fapi\u002Fv1\";\r\n    config.modelId = \"Qwen3-4B-Instruct-2507-GGUF\";\r\n\r\n    gaia::Agent agent(config);\r\n    agent.processQuery(\"Check system health\");\r\n    return 0;\r\n}\r\n```\r\n\r\n**Build requirements:** CMake 3.14+, C++17 compiler. All dependencies fetched automatically via `FetchContent` (`nlohmann\u002Fjson` 3.11.3, `cpp-httplib` 0.15.3, `googletest` 1.14.0).\r\n\r\n```bash\r\ncd cpp\r\ncmake -B build -DCMAKE_BUILD_TYPE=Release\r\ncmake --build build --config Release\r\n```\r\n\r\nSee the [C++ Framework documentation](\u002Fcpp\u002Fquickstart) for the full setup guide and API reference.\r\n\r\n### Health Agent (C++)\r\n\r\nA Windows system health agent that uses the MCP Windows server for PowerShell-based diagnostics. Two modes of operation:\r\n\r\n| Mode | Description |\r\n|------|-------------|\r\n| Quick summary (option 1) | Console-only output — 4 key metrics, no Notepad |\r\n| Full diagnostics (option 15) | Comprehensive report with 12+ tool calls, written to file and opened in Notepad |\r\n\r\nUses direct file-write + `Start-Process notepad` approach for reliable report delivery. Runs with 32K context to support comprehensive multi-tool diagnostics.\r\n\r\n### WiFi Agent (C++)\r\n\r\nA network diagnostic agent that performs WiFi troubleshooting through MCP tool calls. Includes shell injection prevention and safe argument validation.\r\n\r\n### CleanConsole (C++)\r\n\r\nNew `CleanConsole` class providing a polished terminal UI for C++ agents with ANSI color output and automatic word-wrapping — a native equivalent of Python's `AgentConsole`.\r\n\r\n### C++ CI\u002FCD Pipeline\r\n\r\nFull cross-platform CI via `.github\u002Fworkflows\u002Fbuild_cpp.yml`:\r\n- **Cloud CI:** Ubuntu + Windows builds with mock unit tests, install verification, shared library builds\r\n- **STX hardware integration tests:** LLM, MCP, WiFi, and Health agent tests on real AMD hardware\r\n- Uses `Qwen3-4B-Instruct-2507-GGUF` model for integration testing\r\n- Sequential test execution (`-j 1`) to prevent LLM server contention\r\n\r\n---\r\n\r\n## Security\r\n\r\n### Dependabot & Code Scanning Fixes\r\n\r\nResolved **37 security alerts** across the codebase (PR #352):\r\n\r\n**Dependabot (15 alerts):**\r\n\r\n| Package | Severity | Fix |\r\n|---------|----------|-----|\r\n| `tar` | High | Force `>=7.5.8` — fixes path traversal and symlink vulnerabilities |\r\n| `qs` | Low | Update to 6.14.2 — fixes arrayLimit bypass DoS |\r\n| `lodash` | Medium | Update to 4.17.23 — fixes prototype pollution |\r\n\r\n**Code Scanning (22 alerts):**\r\n\r\n| Category | Count | Fix |\r\n|----------|-------|-----|\r\n| Stack trace exposure | 6 | Replace `str(e)` in HTTP responses with generic messages |\r\n| Path injection | 8 | Add path traversal validation and `..` checks |\r\n| Clear-text logging | 4 | Remove patient IDs from log messages |\r\n| Missing workflow permissions | 2 | Add `permissions: contents: read` to CI jobs |\r\n| URL redirect | 1 | Use explicit HTTP 303 redirect status |\r\n| Missing rate limiting | 1 | Add `express-rate-limit` to login endpoint |\r\n\r\n---\r\n\r\n##","2026-03-07T02:57:50",{"id":203,"version":204,"summary_zh":205,"released_at":206},71133,"v0.15.4.1","## What's Changed\n* Release v0.15.4.1 (#350) (73349b9)\n* Fix gaia sd terminal preview and image viewer (#346) (8d12a4a)\n* Fix gaia talk: mic sensitivity, LEMONADE_BASE_URL, stuck listening (#347) (#348) (1198af5)\n* Refine MCP Client architecture diagram (#342) (05b6fda)\n* Fix gaia init for remote Lemonade Server (#345) (12acbab)\n* Fix gaia talk 'No module named pip' error (#344) (a094149)\n* Fix MCP time server example to use mcp-server-time (#339) (d26b7a0)\n* Add VLM profile and structured extraction API (#336) (b882930)\n\n## Installation\nInstall GAIA using pip:\n```bash\npip install amd-gaia\n```\n\nOr using uv:\n```bash\nuv pip install amd-gaia\n```\n","2026-02-24T20:03:42",{"id":208,"version":209,"summary_zh":210,"released_at":211},71134,"v0.15.4","## What's Changed\n* Add release notes for v0.15.4 and bump version (#337) (8e98962)\n* MCP Client Support: Connect GAIA agents to any MCP server (#277) (290fb13)\n* Update VLM model to Qwen3-VL-4B-Instruct-GGUF (#226) (3295114)\n* Fix PowerShell $PID variable conflict in start-lemonade.ps1 (#331) (b209074)\n* Add one-command installer scripts for Windows and Linux (#305) (514ba5e)\n* Add Vision SDK to Q2 2026 Roadmap (#326) (0af6fa2)\n* Fix VLM and Chat Documentation Discrepancies (#328) (66116fa)\n* Add option for stx-test machine (#308) (4a8de23)\n* Update eval framework to use SummarizerAgent (#269) (2cca205)\n\n## Installation\nInstall GAIA using pip:\n```bash\npip install amd-gaia\n```\n\nOr using uv:\n```bash\nuv pip install amd-gaia\n```\n","2026-02-10T23:24:13",{"id":213,"version":214,"summary_zh":215,"released_at":216},71135,"v0.15.3.2","## What's Changed\n* Fix missing packages, broken entry points, and add packaging CI tests (#309) (1d75142)\n* Revert \"Add stx-test to runner pool for testing new runners\" (#307) (a5f653c)\n* Add stx-test to runner pool for testing new runners (#306) (e4cb967)\n* Added release notes for v0.15.3.1 patch (#304) (c734837)\n\n## Installation\nInstall GAIA using pip:\n```bash\npip install amd-gaia\n```\n\nOr using uv:\n```bash\nuv pip install amd-gaia\n```\n","2026-02-06T00:15:42",{"id":218,"version":219,"summary_zh":220,"released_at":221},71136,"v0.15.3.1","\r\n# GAIA v0.15.3.1 Release Notes\r\n\r\n**Patch release** with Linux installation improvements and minimal profile optimization.\r\n\r\n---\r\n\r\n## 🎯 Key Changes\r\n\r\n### Minimal Profile Optimization\r\n- **Smaller, faster setup**: Minimal profile now uses `Qwen3-0.6B-GGUF` (400 MB) instead of 4B model (2.5 GB)\r\n- **6x smaller download** for quick starts and testing\r\n- Updated all documentation to reflect new size\r\n\r\n### Linux Installation Improvements\r\n- **Reliable apt install**: Runs `apt update` before install to prevent 404 errors from stale package cache\r\n- **Cleaner help output**: `gaia init -h` now shows only relevant options (removed global parameters like `--use-claude`, `--model`, `--trace`)\r\n- **Better error UX**: Incomplete model downloads now suggest using `--force-models` flag to redownload from scratch\r\n\r\n### Code Quality & Cleanup\r\n- Simplified Linux version check function (34 lines vs 51 lines, -33%)\r\n- Removed unnecessary PATH manipulation on Linux\r\n- Removed debugging code from dpkg installation\r\n- Fixed Pylint errors in lemonade_installer.py\r\n\r\n---\r\n\r\n## 📦 Installation\r\n\r\n```bash\r\npip install --upgrade amd-gaia\r\n```\r\n\r\n**Linux users** (avoid large CUDA dependencies):\r\n```bash\r\npip install --upgrade amd-gaia --extra-index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcpu\r\n```\r\n\r\n---\r\n\r\n## 🐛 Bug Fixes\r\n\r\n- Fixed apt 404 errors on Linux by updating package cache before install\r\n- Fixed conditional logging level setup to support commands without parent_parser\r\n\r\n---\r\n\r\n## 📝 Full Changelog\r\n\r\nSee [PR #302](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fpull\u002F302) for complete details.","2026-02-05T01:14:13",{"id":223,"version":224,"summary_zh":225,"released_at":226},71137,"v0.15.3","# GAIA v0.15.3 Release Notes\r\n\r\n\r\n## Overview\r\n\r\nThis release introduces **Stable Diffusion image generation** with the new SD Agent, **multi-step workflow parameter passing**, and **composable system prompts architecture**. Includes Lemonade 9.2.0 support, comprehensive playbook, and enhanced agent reliability.\r\n\r\n**TL;DR:**\r\n- **New: SD Agent** - Multi-modal image generation + story creation\r\n- **New: SDToolsMixin & VLMToolsMixin** - Add image\u002Fvision capabilities to any agent\r\n- **Fixed: Multi-step workflows** - Agents pass results between steps automatically\r\n- **Improved: Agent reliability** - Smarter loop detection, 16K context\r\n\r\n---\r\n\r\n## What's New\r\n\r\n### SD Agent: Multi-Modal Image Generation\r\n\r\nNew agent demonstrating how to combine image generation with vision analysis for creative workflows. Shows developers how to build multi-modal applications using GAIA's mixin pattern.\r\n\r\n```bash\r\ngaia init --profile sd\r\ngaia sd \"create a robot exploring ancient ruins\"\r\n# LLM enhances prompt → SD generates image (17s) → VLM creates story (15s)\r\n```\r\n\r\n**What you get:**\r\n- **4 SD Models**: SDXL-Base-1.0 (photorealistic), SDXL-Turbo (fast), SD-1.5, SD-Turbo\r\n- **LLM-Enhanced Prompts**: Research-backed keyword strategies automatically applied\r\n- **Vision Analysis**: Image descriptions and Q&A using Vision LLM\r\n- **Story Creation**: Creative narratives generated from images\r\n- **Story Persistence**: Stories saved as `.txt` files alongside images\r\n- **Random Seeds**: Each generation unique by default (specify seed for reproducibility)\r\n\r\n**Performance (AMD Ryzen AI):**\r\n- Image generation: ~17s (SDXL-Turbo, 512x512)\r\n- Story creation: ~15s (Qwen3-VL-4B)\r\n- Total workflow: ~35s\r\n\r\n**Why this helps:** Build creative AI applications (content generation, game assets, storyboarding) without cloud dependencies. Learn multi-modal agent composition in working code.\r\n\r\n**Example implementation:**\r\n```python\r\nfrom gaia.agents.base import Agent\r\nfrom gaia.sd import SDToolsMixin\r\nfrom gaia.vlm import VLMToolsMixin\r\n\r\nclass ImageStoryAgent(Agent, SDToolsMixin, VLMToolsMixin):\r\n    def __init__(self):\r\n        super().__init__(model_id=\"Qwen3-8B-GGUF\")\r\n        self.init_sd(default_model=\"SDXL-Turbo\")  # 3 SD tools\r\n        self.init_vlm()                            # 2 VLM tools\r\n```\r\n\r\nSee [SD Agent Playbook](https:\u002F\u002Famd-gaia.ai\u002Fplaybooks\u002Fsd-agent) for complete tutorial, and [SD User Guide](https:\u002F\u002Famd-gaia.ai\u002Fguides\u002Fsd) for CLI reference.\r\n\r\n### SDToolsMixin: Stable Diffusion SDK\r\n\r\nNew mixin for adding image generation to any agent.\r\n\r\n**How it helps:** Add professional image generation to any agent in 3 lines. Auto-configures optimal settings per model.\r\n\r\n**Features:**\r\n- **4 Models Supported**: SDXL-Base-1.0, SDXL-Turbo, SD-1.5, SD-Turbo\r\n- **3 Auto-registered Tools**: `generate_image()`, `list_sd_models()`, `get_generation_history()`\r\n- **Model-Specific Defaults**: Automatic size, steps, CFG scale per model (e.g., SDXL-Turbo: 512x512, 4 steps, CFG 1.0)\r\n- **Session Tracking**: Generation history maintained in `self.sd_generations` list\r\n- **Composable Prompts**: `get_sd_system_prompt()` provides research-backed prompt engineering per model\r\n\r\n**Usage:**\r\n```python\r\nclass ImageAgent(Agent, SDToolsMixin):\r\n    def __init__(self):\r\n        super().__init__()\r\n        self.init_sd(default_model=\"SDXL-Turbo\")\r\n        # 3 tools auto-registered, ready to use\r\n```\r\n\r\n### VLMToolsMixin: Vision Language Model SDK\r\n\r\nNew mixin for adding vision capabilities to any agent.\r\n\r\n**How it helps:** Enable agents to understand and analyze images. Access vision client for building custom vision-based tools.\r\n\r\n**Features:**\r\n- **2 Auto-registered Tools**: `analyze_image()`, `answer_question_about_image()`\r\n- **Multi-Model Support**: Qwen3-VL-4B, Qwen2.5-VL-7B, and other vision models\r\n- **Client Access**: `self.vlm_client.extract_from_image()` for building custom tools\r\n- **Composable Prompts**: `get_vlm_system_prompt()` provides usage guidelines\r\n\r\n**Usage:**\r\n```python\r\nclass VisionAgent(Agent, VLMToolsMixin):\r\n    def __init__(self):\r\n        super().__init__()\r\n        self.init_vlm(model=\"Qwen3-VL-4B-Instruct-GGUF\")\r\n        # 2 tools auto-registered: analyze_image, answer_question_about_image\r\n```\r\n\r\n**Design note:** `create_story_from_image` implemented as custom tool in SDAgent (not in VLMToolsMixin) to demonstrate building specialized tools using `self.vlm_client`. Encourages custom tool development over bloating mixins with every use case.\r\n\r\n### Multi-Step Workflow Parameter Passing\r\n\r\nFramework improvement enabling agents to pass results between steps automatically.\r\n\r\n**How it helps:** Build complex workflows (data fetch → process → analyze → store) without manual result passing. Works for all agents, not just SD Agent.\r\n\r\n**Problem:** Multi-step workflows failed because agents couldn't reference previous outputs. Resulted in \"Image not found\" errors when step 2 needed step 1's image_path.\r\n\r\n**Solution:** Placeholder syntax","2026-02-04T07:48:15",{"id":228,"version":229,"summary_zh":230,"released_at":231},71138,"v0.15.2","## Overview\r\n\r\nThis release focuses on **streamlined setup experience** with improvements to `gaia init`, **Lemonade 9.1.4 compatibility**, expanded **MCP ecosystem roadmap**, and numerous **documentation fixes** based on manual testing.\r\n\r\n## What's New\r\n\r\n### Improved `gaia init` Command\r\nEnhanced one-stop setup command for reliability and user experience:\r\n- **Simplified Download Flow**: Downloads all required models directly instead of pre-checking availability\r\n- **Download Speed Display**: Progress bar now shows download speed (e.g., `@ 75 MB\u002Fs`)\r\n- **Remote Server Support**: New `--remote` flag for remote Lemonade server setups\r\n- **Graceful Shutdown**: `gaia kill --lemonade` now uses graceful `lemonade-server stop`\r\n- **Removed Redundancy**: `gaia pull` command removed (use `lemonade-server pull` instead)\r\n\r\n```bash\r\n# Quick setup with speed display\r\ngaia init\r\n\r\n# For remote Lemonade server setups\r\ngaia init --remote\r\n```\r\n\r\n### Lemonade 9.1.4 Support\r\nFull compatibility with Lemonade Server 9.1.4:\r\n- **Health Check Updates**: Updated health check format handling\r\n- **Version Display**: Docs navbar shows `v0.15.2 · Lemonade 9.1.4`\r\n\r\n### Computer Use Agent (CUA) Documentation\r\nNew documentation and roadmap for the Computer Use Agent:\r\n- **Technical Specification**: GUI automation capabilities\r\n- **Roadmap**: Screen capture, element detection, action execution\r\n- **Integration Patterns**: Works with existing agent framework\r\n\r\n### MCP Ecosystem Roadmap\r\nExpanded Model Context Protocol plans for Q1 2026:\r\n- **MCP Client Mixin**: Plan for client-side MCP integration\r\n- **Ecosystem Roadmap**: Comprehensive Q1 2026 timeline\r\n- **Documentation**: Enhanced MCP documentation coverage\r\n\r\n### Release Branch Automation\r\nGitHub Action automatically updates the `release` branch on tag push:\r\n- **Mintlify Integration**: Seamless docs deployment on releases\r\n- **Version Tracking**: Navbar displays current version automatically\r\n\r\n> ⚠️ **Breaking Change**: The `gaia pull` command has been removed. Use `lemonade-server pull` directly instead.\r\n\r\n## Improvements\r\n\r\n### Developer Experience\r\n- **Lint Script Improvements**: Enhanced `--fix` mode for better auto-formatting (#229)\r\n- **uvx Auto-Download**: Lint utilities now use uvx for automatic tool installation (#218)\r\n- **Import Validation**: Fixed import inconsistencies and enhanced validation (#204)\r\n- **Max-Steps Warning**: Now appears after final step completes, not before (#249)\r\n\r\n### Documentation\r\nBased on manual testing, several guides were corrected:\r\n- **Chat SDK Guide**: Fixed examples and workflows (#194)\r\n- **Blender Agent Guide**: Corrected setup and usage instructions (#195)\r\n- **Hardware Advisor Playbook**: Fixed structure and renamed to index.mdx (#217)\r\n- **Hardware Advisor Agent**: Fixed 'No Plan Found' warning for simple queries (#216)\r\n\r\n### UI\u002FUX\r\n- **Navbar Improvements**: Added bottom border and improved tab visibility (#212)\r\n\r\n### Infrastructure\r\n- **Legacy Installer Removed**: NSIS Windows Installer removed in favor of Python packages (#192)\r\n- **Dependabot**: Set to monitor-only mode to reduce PR noise\r\n- **GitHub Actions**: Bumped action versions across the repository (#198)\r\n\r\n## Bug Fixes\r\n\r\n- **#249**: Max-steps warning timing - now appears after final step completes\r\n- **#221**: Related max-steps warning timing issue\r\n- **#250**: Model download flow reliability in `gaia init`\r\n- **#216**: Hardware Advisor Agent 'No Plan Found' warning for simple queries\r\n\r\n## Breaking Changes\r\n\r\n| Change | Migration |\r\n|--------|-----------|\r\n| `gaia pull` removed | Use `lemonade-server pull` directly |\r\n| Legacy NSIS installer removed | Use Python package installation (`pip install amd-gaia`) |\r\n\r\n## Full Changelog\r\n\r\n**18 commits** from multiple contributors\r\n\r\nKey PRs:\r\n- #249 - Fix max-steps warning timing to appear after final step completes\r\n- #219 - Add `gaia init` command for one-stop setup\r\n- #229 - Improve `--fix` mode for lint scripts\r\n- #228 - Update and support Lemonade 9.1.4 health check format\r\n- #225 - Add Computer Use Agent (CUA) Documentation and Roadmap\r\n- #218 - Update lint utilities to use uvx for auto-downloading tools\r\n- #217 - Fix Hardware Advisor Playbook structure\r\n- #216 - Fix Hardware Advisor Agent 'No Plan Found' warning\r\n- #212 - Add navbar bottom border and improve tab visibility\r\n- #206 - MCP Client Mixin Plan and Roadmap\r\n- #204 - Fix Import Inconsistencies and Enhance Validation\r\n- #202 - Q1 2026 MCP Ecosystem Roadmap\r\n- #198 - Bump the github-actions group\r\n- #195 - Fix Blender Agent guide based on manual testing\r\n- #194 - Fix Chat SDK guide based on manual testing\r\n- #192 - Remove Legacy NSIS Windows Installer\r\n\r\nFull Changelog: [v0.15.1...v0.15.2](https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.15.1...v0.15.2)\r\n","2026-01-24T04:46:59",{"id":233,"version":234,"summary_zh":235,"released_at":236},71139,"v0.15.1","# GAIA v0.15.1 Release Notes\r\n\r\n## Overview\r\n\r\nThis release introduces the **Summarization Agent** with MCP integration, refactors the **LLM Client Architecture** to a provider-based pattern, adds **Claude AI Assistant automation** for GitHub issues and PRs, and significantly improves **developer tooling** with automated release notes, simplified CLI, and new evaluation benchmarks.\r\n\r\n## Installation\r\n\r\n```bash\r\n# Install uv (ultra-fast Python package manager)\r\n# Windows: irm https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.ps1 | iex\r\n# macOS\u002FLinux: curl -LsSf https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.sh | sh\r\n\r\ngit clone https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia.git\r\ncd gaia\r\nuv venv .venv --python 3.12\r\nsource .venv\u002Fbin\u002Factivate  # Windows: .\\.venv\\Scripts\\Activate.ps1\r\nuv pip install -e .\r\ngaia -v\r\n```\r\n\r\nOr install from PyPI:\r\n```bash\r\nuv pip install amd-gaia\r\n```\r\n\r\n## What's New\r\n\r\n### 📝 Summarization Agent\r\nNew agent for document summarization with MCP bridge integration:\r\n- **MCP Bridge Integration**: Data sent as `multipart\u002Fform-data` leveraging OCR capabilities\r\n- **Streaming Responses**: Summarization results streamed in real time\r\n- **Iterative Summarization**: Produced iteratively to minimize time-to-first-token (TTFT)\r\n- **PDF Text Caching**: Extracted text cached for improved performance\r\n- **KV Cache Optimization**: Reduced TTFT by leveraging KV cache properly\r\n\r\n```bash\r\ngaia summarize document.pdf\r\n```\r\n\r\n### 🔧 LLM Client Factory\r\nComplete refactor of LLM client architecture for better maintainability:\r\n- **Provider Pattern**: New `LemonadeProvider`, `OpenAIProvider`, `ClaudeProvider` implementations\r\n- **Factory Function**: Easy instantiation with `create_client(\"lemonade\")`\r\n- **Base Interface**: Abstract `LLMClient` class for consistent behavior\r\n- **Auto-Loading**: Models automatically load before requests in `LemonadeClient`\r\n- **Default Temperature**: 0.1 for deterministic responses\r\n\r\n```python\r\nfrom gaia.llm import LLMClient, create_client\r\n\r\n# New pattern\r\nclient = create_client(\"lemonade\")  # or \"openai\", \"claude\"\r\nresponse = client.chat(\"Hello!\")\r\n```\r\n\r\n### 🤖 Claude AI Assistant Workflow\r\nAutomated Claude assistance for GitHub issues and pull requests:\r\n- **PR Review**: Automatic reviews on PR open\u002Fready with security and AMD compliance checks\r\n- **Issue Handling**: Intelligent triage and response to new issues\r\n- **@claude Mentions**: Responds to mentions in PR comments and issues\r\n- **Documentation-Aware**: References docs\u002Fdocs.json for intelligent responses\r\n- **Cost-Optimized**: Concurrency control and selective triggering\r\n\r\n### 📋 Automated Release Notes Generation\r\nClaude-powered release notes when GitHub releases are created:\r\n- **Dual Output**: `RELEASE_NOTES.md` (GitHub) + `docs\u002Freleases\u002FvX.Y.Z.mdx` (website)\r\n- **Iterative Diff Analysis**: Splits diffs by component for large releases\r\n- **Auto Version Bump**: Automatically bumps to next patch version after release\r\n- **MDX Validation**: Validates frontmatter, required sections, and changelog links\r\n- **Self-Review**: Claude verifies its generated output before committing\r\n\r\n### 💻 GAIA Code CLI Simplification\r\nStreamlined CLI by removing unnecessary subcommands:\r\n- **Direct Invocation**: `gaia-code \"Build me an app\"` works directly\r\n- **Auto-Initialization**: Models load automatically on first use\r\n- **176 Lines Removed**: Cleaner, simpler codebase\r\n- **Working Help**: `gaia-code --help` now shows ALL arguments\r\n\r\n```bash\r\n# These all work directly now:\r\ngaia-code \"Build me a todo app\"\r\ngaia-code \"Build me an app\" --path ~\u002Fprojects\u002Fmyapp\r\ngaia-code --interactive\r\n```\r\n\r\n**Breaking Change**: The `run` subcommand has been removed:\r\n- ❌ `gaia-code run \"Build me an app\"` (no longer works)\r\n- ✅ `gaia-code \"Build me an app\"` (new syntax)\r\n\r\n### 🧪 Fix-Code Microbenchmark\r\nNew evaluation framework for automated code fixes:\r\n- **CLI Helper**: `gaia eval fix-code` command\r\n- **Prompt Engineering**: Experiment with prompt designs for code fixes\r\n- **Multi-Model Support**: Test with local models or Claude\r\n- **Sample Fixtures**: Python and TypeScript bug examples included\r\n- **Diff Output**: Shows patched output with diffs\r\n\r\n```bash\r\ngaia eval fix-code --model claude examples\u002Fsum.py\r\n```\r\n\r\n### 📊 Performance Analysis Plotter\r\nNew CLI tool for analyzing LLM server performance:\r\n- **Log Analysis**: Ingests llama.cpp server logs\r\n- **Token Charts**: Generates prompt\u002Finput\u002Foutput token count charts\r\n- **Performance Metrics**: TTFT and TPS plots\r\n- **Prefill vs Decode**: Pie charts showing time distribution\r\n\r\n```bash\r\ngaia perf-analysis --show server.log\r\n```\r\n\r\n### 🗺️ Public Roadmap\r\nNew documentation section with transparent development plans:\r\n- **Roadmap Page**: Timeline and upcoming priorities at [amd-gaia.ai](https:\u002F\u002Famd-gaia.ai)\r\n- **Technical Plans**: Detailed specs for Chat UI and Installer\r\n- **Q1 2026 Timeline**: Visual Mermaid diagram showing planned features\r\n- **Community Engagement**: Email contact (gaia@amd.com) for use cases\r\n\r\n## Improvements\r\n\r\n### Claude Frame","2026-01-17T02:27:19",{"id":238,"version":239,"summary_zh":240,"released_at":241},71140,"v0.15.0","# GAIA v0.15.0 Release Notes\r\n\r\n## Overview\r\n\r\nThis release transforms GAIA into a full **AI Agent Framework (SDK v1.0.0)**, introduces the **Medical Intake Agent with Dashboard**, adds the **Database Module** for SQLite-backed agents, and includes comprehensive documentation improvements with new playbooks.\r\n\r\n## Installation\r\n\r\n```bash\r\n# Install uv (ultra-fast Python package manager)\r\n# Windows: irm https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.ps1 | iex\r\n# macOS\u002FLinux: curl -LsSf https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.sh | sh\r\n\r\ngit clone https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia.git\r\ncd gaia\r\nuv venv .venv --python 3.12\r\nsource .venv\u002Fbin\u002Factivate  # Windows: .\\.venv\\Scripts\\Activate.ps1\r\nuv pip install -e .\r\ngaia -v\r\n```\r\n\r\nOr install from PyPI:\r\n```bash\r\nuv pip install amd-gaia\r\n```\r\n\r\n## What's New\r\n\r\n### 🚀 SDK v1.0.0 - AI Agent Framework\r\nGAIA is now positioned as a **pure framework\u002FSDK** for building AI PC agents:\r\n- 900+ line API reference with 20+ components\r\n- 60+ MDX documentation files organized into tabs\r\n- 55+ interface validation tests\r\n- Mintlify-powered documentation site at [amd-gaia.ai](https:\u002F\u002Famd-gaia.ai)\r\n\r\n```python\r\nfrom gaia import Agent, tool\r\n\r\nclass MyAgent(Agent):\r\n    def _get_system_prompt(self) -> str:\r\n        return \"You are a helpful assistant.\"\r\n\r\n    def _register_tools(self):\r\n        @tool\r\n        def greet(name: str) -> str:\r\n            \"\"\"Greet someone by name.\"\"\"\r\n            return f\"Hello, {name}!\"\r\n```\r\n\r\n### 🏥 Medical Intake Agent with Dashboard\r\nComplete patient intake form processing system:\r\n- Automatic file watching for intake forms (.png, .jpg, .pdf)\r\n- VLM-powered data extraction (Qwen2.5-VL on NPU)\r\n- SQLite database with 17 patient fields\r\n- Real-time React dashboard with SSE updates\r\n- Natural language patient queries\r\n\r\n```bash\r\ngaia-emr watch          # Auto-process forms\r\ngaia-emr dashboard      # Launch web dashboard\r\ngaia-emr query \"Find patient John Smith\"\r\n```\r\n\r\n### 🗄️ Database Module\r\nNew `gaia.database` module with two usage patterns:\r\n\r\n**DatabaseAgent** (prototyping):\r\n```python\r\nfrom gaia import DatabaseAgent\r\n\r\nclass MyAgent(DatabaseAgent):\r\n    def __init__(self, **kwargs):\r\n        super().__init__(db_path=\"data\u002Fapp.db\", **kwargs)\r\n```\r\n\r\n**DatabaseMixin** (production):\r\n```python\r\nfrom gaia import Agent, DatabaseMixin, tool\r\n\r\nclass MyAgent(Agent, DatabaseMixin):\r\n    def __init__(self, **kwargs):\r\n        super().__init__(**kwargs)\r\n        self.init_db(\"data\u002Fapp.db\")\r\n```\r\n\r\n### 🧪 Testing Utilities Module\r\nNew `gaia.testing` module for testing agents without real LLM\u002FVLM services:\r\n- `MockLLMProvider`, `MockVLMClient`, `MockToolExecutor`\r\n- `temp_directory`, `temp_file`, `create_test_agent` fixtures\r\n- `assert_llm_called`, `assert_tool_called` assertions\r\n- 51 unit tests with full coverage\r\n\r\n```python\r\nfrom gaia.testing import MockLLMProvider, assert_llm_called\r\n\r\ndef test_my_agent():\r\n    mock_llm = MockLLMProvider(responses=[\"I found the data\"])\r\n    agent = MyAgent(skip_lemonade=True)\r\n    agent.chat = mock_llm\r\n    result = agent.process_query(\"Find data\")\r\n    assert_llm_called(mock_llm)\r\n```\r\n\r\n### 💻 Hardware Advisor Agent\r\nNew example agent with LemonadeClient APIs for dynamic model recommendations:\r\n```bash\r\npython examples\u002Fhardware_advisor_agent.py\r\n```\r\n\r\n### 📚 GAIA Code Playbook\r\nComplete 3-part playbook for the Code Agent:\r\n- Part 1: Introduction and fundamentals\r\n- Part 2: Application creation and development\r\n- Part 3: Validation and building workflows\r\n\r\n## Improvements\r\n\r\n### Agent UX Enhancements\r\n- **Rich Panel Final Answers**: Agent responses now appear with green-bordered panels and \"✅ Final Answer\" title\r\n- **Better Completion Messages**: Removed confusing step counters, now shows clean \"Processing complete!\" message\r\n- **Enhanced Error Display**: Shows execution trace with Query → Plan Step → Tool → Error, plus code context with line pointers\r\n\r\n### Code Agent\r\n- Limited router to TypeScript only (other languages report helpful error)\r\n- Fixed default path issue where `create-next-app` would fail on non-empty directories\r\n- Added dynamic timer scaffolding and artifact-aware planning\r\n\r\n### Lemonade Server Integration\r\n- Centralized initialization in `LemonadeManager` singleton\r\n- Changed context size check from error to warning (agents continue running)\r\n- Fixed VLM image processing breaking embeddings endpoint\r\n\r\n### Documentation\r\n- New `setup.mdx` with step-by-step installation guide\r\n- New `glossary.mdx` with 50+ GAIA terms\r\n- Streamlined learning path: Quickstart → Hardware Advisor → Playbooks\u002FGuides\r\n- Applied `\u003CCodeGroup>` component across multiple docs for compact command examples\r\n- Standardized license headers across 90+ files\r\n- Added source code links throughout documentation\r\n\r\n### PyPI Package\r\n- Renamed from `gaia` to `amd-gaia` (import name remains `gaia`)\r\n- Enhanced PyTorch CPU-only build support\r\n\r\n## Bug Fixes\r\n\r\n- **#1092**: Agent receives empty response after tool execution - fixed duplicate message insertion\r\n- **#1088**: Con","2026-01-06T02:07:20",{"id":243,"version":244,"summary_zh":245,"released_at":246},71141,"v0.14.1","# GAIA v0.14.1 Release Notes\r\n\r\n## Overview\r\n\r\nThis release enhances the **Code Agent** with checklist-based orchestration for web development, upgrades to **Lemonade Server v9.1.0**, and fixes chat history persistence.\r\n\r\n## Installation\r\n\r\n```bash\r\n# Install uv (ultra-fast Python package manager)\r\n# Windows: irm https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.ps1 | iex\r\n# macOS\u002FLinux: curl -LsSf https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.sh | sh\r\n\r\ngit clone https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia.git\r\ncd gaia\r\nuv venv .venv --python 3.12\r\nsource .venv\u002Fbin\u002Factivate  # Windows: .\\.venv\\Scripts\\Activate.ps1\r\nuv pip install -e .\r\ngaia -v\r\n```\r\n\r\n> **Note:** As GAIA is upgraded, the above flow is the recommended one. A new installer is coming in a future.\r\n## What's New\r\n\r\n### 🛠️ Code Agent Enhancements\r\n\r\nNew orchestration framework for building web applications:\r\n\r\n- Checklist-driven workflows that break complex tasks into structured steps\r\n- Automatic project type detection (Next.js, Python) with appropriate tooling\r\n- Conversation history summarization for faster debugging cycles\r\n- Validation tools for build, lint, and type-checking\r\n\r\n```bash\r\ngaia code \"Create a task management app with user authentication\"\r\n```\r\n\r\n### 🍋 Lemonade Server v9.1.0\r\n\r\n- Upgraded to Lemonade Server v9.1.0\r\n- Health check verifies Lemonade installation with clear error messages if missing\r\n- Context size validation ensures sufficient tokens before agent execution\r\n\r\n### 💬 Chat Improvements\r\n\r\n- **History persistence fix**: Conversation history now properly saves with `\u002Fsave` and restores with `\u002Fresume`\r\n- **Better no-document behavior**: Chat agent uses general knowledge instead of failing when no documents are indexed\r\n\r\n## Improvements\r\n\r\n- **Linting:** Cross-platform linting script (`util\u002Flint.py`) for Windows\u002FmacOS\u002FLinux\r\n- **CI\u002FCD:** New chat agent test workflow\r\n\r\n## What's Changed\r\n\r\n- GAIA Code Enhancements by @itomek\r\n- Implement cross-platform linting script by @kovtcharov\r\n- Add health check for Lemonade server installation by @kovtcharov\r\n- Add context size validation for Lemonade server by @kovtcharov\r\n- Fix Chat History Persistence by @kovtcharov\r\n- Enhance chat agent behavior when no documents indexed by @kovtcharov\r\n- Faster Web Dev Agent Iterations by @eddierichter-amd\r\n- Update Lemonade version to 9.1.0 by @kovtcharov\r\n\r\n**Full Changelog:** https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.14.0...v0.14.1","2025-12-15T23:31:34",{"id":248,"version":249,"summary_zh":250,"released_at":251},71142,"v0.14.0","# GAIA v0.14.0 Release Notes\r\n\r\n## Overview\r\n\r\nThis release introduces the **Knowledge Assistant** for document Q&A with agentic RAG, transitions to a streamlined **cross-platform developer workflow** using `uv`, and includes Lemonade Server v9.0.8 with auto-start capabilities.\r\n\r\n## Installation\r\n```bash\r\n# Install uv (ultra-fast Python package manager)\r\n# Windows: irm https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.ps1 | iex\r\n# macOS\u002FLinux: curl -LsSf https:\u002F\u002Fastral.sh\u002Fuv\u002Finstall.sh | sh\r\n\r\ngit clone https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia.git\r\ncd gaia\r\nuv venv .venv --python 3.12\r\nsource .venv\u002Fbin\u002Factivate  # Windows: .\\.venv\\Scripts\\Activate.ps1\r\nuv pip install -e .\r\ngaia -v\r\n```\r\n\r\n**Why uv?** 10-100x faster installs, automatic Python management, cross-platform support (Windows\u002FmacOS\u002FLinux), and editable installs.\r\n\r\n> **Note:** As GAIA is upgraded, the above flow is the recommended one. A new installer is coming in a future.\r\n\r\n## What's New\r\n\r\n### 📚 Knowledge Assistant - Document Q&A\r\n\r\nChat with your documents using agentic RAG:\r\n\r\n- Index PDFs, markdown, text, CSV, JSON, and 30+ code file types\r\n- Semantic search with hybrid keyword boosting\r\n- VLM image extraction from PDFs\r\n- Auto-discovery: agent searches, indexes, and answers automatically\r\n```bash\r\ngaia chat --index manual.pdf\r\ngaia rag quick report.pdf \"What are the key findings?\"\r\ngaia talk --index document.pdf\r\n```\r\n\r\n### 🍎 Cross-Platform Support\r\n\r\n- Native macOS and Linux support\r\n- Removed Conda dependency, migrated to Python venv\r\n- macOS testing in CI\u002FCD\r\n\r\n### 🍋 Lemonade Server v9.0.8 Auto-Start\r\n\r\n- GAIA automatically starts Lemonade Server if not running\r\n- Version compatibility checks\r\n- `--base-url` support for custom endpoints\r\n\r\n### ⬇️ Model Pull with Streaming Progress\r\n```bash\r\ngaia model pull Qwen2.5-3B-Instruct-GGUF\r\n```\r\n\r\nDownload models with real-time progress updates and resume support.\r\n\r\n### 🛠️ Code Agent Improvements\r\n\r\n- Enhanced debugging capabilities\r\n- Structured tool role messages\r\n- Bug fixes and deprecated tool cleanup\r\n\r\n### 🔗 Unified --base-url CLI Support\r\n\r\nUse custom Lemonade Server URLs across all commands:\r\n```bash\r\ngaia chat --base-url http:\u002F\u002Fcustom-server:8000\r\n```\r\n\r\n## Improvements\r\n\r\n- **Evaluation:** Transcript validation, extended timeouts, resume\u002Fretry for groundtruth\r\n- **Security:** Path traversal prevention with PathValidator\r\n- **Infrastructure:** Constants refactoring, localhost reference updates\r\n- **Developer Experience:** API documentation, Lemonade MSI installer, Node.js v20 VSCode prereq\r\n- **CI\u002FCD:** Workflow updates\r\n\r\n**Full Changelog:** https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.13.0...v0.14.0","2025-12-11T14:45:16",{"id":253,"version":254,"summary_zh":255,"released_at":256},71143,"v0.13.0","# GAIA v0.13.0 Release Notes\r\n\r\n## Overview\r\n\r\nThis major release introduces **GAIA Code**, a proof-of-concept AI coding agent with VSCode integration, and a **Docker Agent** for containerized workflows. Most significantly, this release establishes a new architecture that allows GAIA agents to be easily built, and to be exposed via **API, MCP, and CLI**, opening up extensive possibilities for agent composition and integration. The release also includes improvements to the evaluation framework and enhanced documentation for building custom agents.\r\n\r\n## What's New\r\n\r\n### 🚀 GAIA Code Agent with VSCode Integration (#774, #864)\r\n\r\nIntroduced a proof-of-concept AI-powered coding agent with Visual Studio Code integration:\r\n\r\n- **In-Editor Development**: Trigger GAIA Code Agent from VSCode via extension coupled with GitHub Copilot\r\n- **Automated Code Generation**: Generate Python code from natural language descriptions\r\n- **Test-Driven Development**: Automatic test generation and execution\r\n- **Iterative Refinement**: Multi-iteration approach to code quality\r\n- **File Context Awareness**: Automatic workspace file monitoring and context\r\n\r\n*Note: GAIA Code is currently a proof-of-concept focused on Python development workflows.*\r\n\r\n**Files Changed**: `src\u002Fgaia\u002Fagents\u002Fcode\u002F`, VSCode extension files\r\n\r\n### 🐳 Docker Agent (#811, #833)\r\n\r\nNew proof-of-concept Docker agent for containerized application development:\r\n\r\n- **Container Management**: Create, start, stop, and manage Docker containers\r\n- **Image Building**: Automated Dockerfile generation and image builds\r\n- **Docker Compose Support**: Multi-container orchestration capabilities\r\n- **Isolated Environments**: Containerized development environments for projects\r\n\r\n*Note: The Docker Agent is currently a proof-of-concept demonstrating containerized workflow automation.*\r\n\r\n### 🏗️ Agent Architecture: Multi-Protocol Exposure (#846)\r\n\r\n**Major architectural enhancement** enabling GAIA agents to be exposed through multiple interfaces:\r\n\r\n- **API Exposure**: RESTful API endpoints for agent interactions\r\n- **MCP (Model Context Protocol)**: Native MCP server support for agent communication\r\n- **CLI Interface**: Command-line access to agent capabilities\r\n- **Unified Pattern**: Class inheritance-based design pattern for building new agents\r\n\r\nThis architecture opens up powerful possibilities:\r\n- Integrate agents into existing tools and workflows\r\n- Build custom agents using established patterns\r\n- Mix and match communication protocols based on needs\r\n\r\n## Improvements\r\n\r\n### 🔧 Evaluation Framework Improvements\r\n\r\n#### Fix: Remove Silent Fallback in Transcript Matching (#843)\r\n\r\nImproved error handling in evaluation transcript matching:\r\n\r\n- **Fail-Fast Approach**: Clear failures instead of silent fallbacks\r\n- **Better Debugging**: Improved error messages for mismatches\r\n- **Data Integrity**: Ensures evaluation data consistency\r\n\r\n#### Restore Python Execution Tools (#839)\r\n\r\nRe-added essential Python execution capabilities:\r\n\r\n- Restored `run_test` tool for test execution\r\n- Restored `execute_python_file` tool for script running\r\n- Better integration with code agent workflows\r\n\r\n## Getting Started with Custom Agents\r\n\r\nWith the new agent architecture, building custom agents is straightforward. Agents can inherit from base classes and automatically gain API, MCP, and CLI exposure. See the updated documentation for examples and best practices.","2025-11-14T16:12:16",{"id":258,"version":259,"summary_zh":260,"released_at":261},71144,"v0.12.1","# GAIA v0.12.1 Release Notes\r\n\r\n## Overview\r\n\r\nThis patch release focuses on bug fixes and improvements to the evaluation framework, particularly addressing issues with the visualization and reporting tools. All changes improve the reliability and usability of the `gaia eval`, `gaia visualize`, and `gaia report` commands.\r\n\r\n## What's Changed\r\n\r\n### Bug Fixes\r\n\r\n#### 🔧 Fix Evaluation Visualizer Model Count and Path Issues (#823)\r\n\r\nFixed multiple critical issues in the `gaia visualize` and `gaia report` commands:\r\n\r\n- **Incorrect Model Count in Consolidated Report**: Fixed model count calculation in the webapp to show the correct number of models (was showing only 4 instead of 8)\r\n  - Now calculates unique models directly from `metadata.evaluation_files` instead of filtered\u002Fgrouped data\r\n\r\n- **Windows Path Separator Bug**: Fixed cross-platform compatibility issue in `isMainEvaluationEntry()` function\r\n  - Now handles both Unix (`\u002F`) and Windows (`\\`) path separators correctly\r\n\r\n- **Incorrect Default Directory Paths**: Updated default paths to match actual evaluation output locations\r\n  - Changed from `workspace\u002Fevaluation` to `workspace\u002Foutput\u002Fevaluations`\r\n  - Changed from `workspace\u002Fexperiments` to `workspace\u002Foutput\u002Fexperiments`\r\n\r\n- **Outdated Report Filename**: Updated default report filename from `LLM_RAG_Evaluation_Report.md` to `LLM_Evaluation_Report.md`\r\n  - Better reflects support for multiple evaluation types (RAG, summarization, etc.)\r\n\r\n**Files Changed**: `src\u002Fgaia\u002Fcli.py`, `src\u002Fgaia\u002Feval\u002Feval.py`, `src\u002Fgaia\u002Feval\u002Fwebapp\u002Fpublic\u002Fapp.js`\r\n\r\n### Improvements\r\n\r\n#### 📊 Standardize Evaluation Workflow Default Directories (#820)\r\n\r\nImplemented consistent default parameters across all evaluation commands with a unified directory structure:\r\n\r\n```\r\n.\u002Foutput\u002F\r\n├── test_data\u002F          # gaia generate\r\n├── groundtruth\u002F        # gaia groundtruth\r\n├── experiments\u002F        # gaia batch-experiment\r\n└── evaluations\u002F        # gaia eval\r\n```\r\n\r\n**Key Changes**:\r\n- Added centralized directory constants in `cli.py`\r\n- Added `GAIA_WORKSPACE` environment variable support for flexible workspace management\r\n- Updated all command defaults to use the new structure\r\n- Updated documentation in `docs\u002Feval.md` and `docs\u002Fcli.md`\r\n\r\n**Benefits**:\r\n- Consistency: All evaluation artifacts organized in one location\r\n- Maintainability: Centralized constants eliminate duplication\r\n- Flexibility: Workspace environment variable for managing multiple projects\r\n- Cleanup: Single directory to clean or ignore\r\n\r\n**Files Changed**: Multiple files including CLI, evaluation modules, webapp components, and documentation\r\n\r\n#### 🏷️ Improve Reporting for Cloud Model Identifiers (#834)\r\n\r\nEnhanced model counting logic in the Evaluation Visualizer to support additional cloud model identifiers:\r\n\r\n- Added support for 'gpt-4' and 'gemini' model identifiers\r\n- Improved accuracy of model classification in reports\r\n\r\n**Files Changed**: `src\u002Fgaia\u002Feval\u002Fwebapp\u002Fpublic\u002Fapp.js`\r\n\r\n## Contributors\r\n\r\n- Kalin Ovtcharov (@kalin-ovtcharov)\r\n\r\n## Upgrade Notes\r\n\r\nIf you have existing evaluation workflows, note the following directory changes:\r\n\r\n- `.\u002Fevaluation` → `.\u002Foutput\u002Fevaluations`\r\n- `.\u002Fexperiments` → `.\u002Foutput\u002Fexperiments`\r\n\r\nYou can set the `GAIA_WORKSPACE` environment variable to use a custom workspace location if needed.\r\n\r\n---\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Famd\u002Fgaia\u002Fcompare\u002Fv0.12.0...v0.12.1\r\n","2025-10-22T06:45:22",{"id":263,"version":264,"summary_zh":265,"released_at":266},71145,"v0.12.0","# Release v0.12.0 - Docker Agent Integration\r\n\r\n## Features & Enhancements\r\n\r\n### Docker Agent Integration (#810, #811)\r\n- Added Docker agent for natural language containerization with AI-powered Dockerfile generation\r\n- Implemented modular MCP architecture with per-agent server support using FastMCP\r\n- Created `gaia mcp docker` command for standalone Docker agent MCP server\r\n- Added Docker application framework for testing and demonstrations\r\n- Enhanced agent system with MCPAgent base class for Model Context Protocol support\r\n\r\n### Model Improvements\r\n- Updated default model to Qwen3-Coder-30B-A3B-Instruct-GGUF for improved code generation performance\r\n- Optimized Dockerfile generation with multi-step planning and validation\r\n\r\n### Architecture Improvements\r\n- Implemented AgentMCPServer for wrapping MCP agents and exposing them via HTTP + JSON-RPC\r\n- Refactored MCP transport layer for better modularity and agent isolation\r\n- Enhanced agent execution with detailed status reporting and result handling\r\n\r\n### Documentation\r\n- Comprehensive Docker agent documentation with setup and usage examples (docker.md)\r\n- Updated MCP documentation with Docker agent integration guide (mcp.md)\r\n- Added CLI examples for Docker agent workflows\r\n\r\n## Demo\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F8d9be153-4f31-4eee-bd30-561b99ee4ba4\r\n","2025-10-17T22:19:38",{"id":268,"version":269,"summary_zh":270,"released_at":271},71146,"v0.11.2","# Release Notes - v0.11.2\r\n\r\n## Features & Enhancements\r\n\r\n### Eval Framework Improvements (#784)\r\n- Added chat template support for better model compatibility\r\n- Implemented thinking token extraction for advanced analysis\r\n- Enhanced batch experiment runner with multi-model configuration support\r\n- Expanded evaluation documentation and webapp UI improvements\r\n- Added new `multi_model_summarization.json` configuration\r\n\r\n## Updates\r\n\r\n### Lemonade Backend (#818)\r\n- Updated to Lemonade v8.1.12\r\n\r\n## Documentation\r\n\r\n### Developer Documentation (#803)\r\n- Clarified repository structure and development workflow\r\n- Updated release process documentation\r\n\r\n## Changes Summary\r\n- 20 files changed, 948 insertions(+), 603 deletions(-)\r\n- Focus areas: evaluation framework enhancements, model compatibility, and documentation clarity\r\n","2025-10-16T03:45:59",{"id":273,"version":274,"summary_zh":275,"released_at":276},71147,"v0.11.1","## Architecture Improvements\r\n\r\n- Code quality improvements — Improved static analysis and cleaned up the codebase\r\n- Static analysis cleanup — Resolved warnings and reduced false positives\r\n\r\n## Testing & CI\u002FCD\r\n\r\n- Lint workflow — Expanded CI to run comprehensive linting, type checks, security scans, and import smoke tests\r\n- Local dev tooling — Added a helper script (util\u002Flint.ps1) to run and optionally fix checks locally\r\n\r\n## Bug Fixes\r\n\r\n- Filesystem — Prevent unintended directory creation when no new directory is required\r\n\r\n## Documentation\r\n\r\n- Model installation — Clarified how to install\u002Fmanage additional models via Lemonade model manager; cross-linked in CLI guide and FAQ\r\n\r\n**Full Changelog**: v0.11.0...v0.11.1","2025-10-10T12:44:41",{"id":278,"version":279,"summary_zh":280,"released_at":281},71148,"v0.11.0","### App Development Framework\r\n\r\nElectron-based framework for building AI-powered desktop applications.\r\n\r\n- Example app template — Ready-to-use MCP integration demo\r\n- NPM integration — Streamlined development workflows\r\n- CI\u002FCD automation — GitHub workflows for building and packaging\r\n\r\n### JAX Desktop Application\r\n\r\nElectron-based Jira Dashboard with integrated AI assistant.\r\n\r\n- Desktop App — Projects, issues, search, and creation\r\n- System status — Real-time GAIA and MCP Bridge monitoring\r\n- AI chat assistant — Context-aware help for Jira workflows\r\n\r\n### n8n Workflow Integration\r\n\r\nComplete integration guide with pre-built workflow templates.\r\n\r\n- Pre-built templates — Common automation scenarios\r\n- HTTP integration — Simple REST API calls to MCP server\r\n- Example workflows — Email summarization, Jira automation, content\r\ngeneration\r\n\r\n## Architecture Improvements\r\n\r\n- Enhanced agent system — Improved state management and tool registry\r\n- Blender agent refactoring — Package renamed to lowercase\r\nagents\u002Fblender\u002F for consistency\r\n- Streaming support — Real-time response streaming throughout agent\r\ninteractions\r\n\r\n## Testing & CI\u002FCD\r\n\r\n- Jira agent tests — Complete test suite with interactive mode\r\n- MCP integration tests — Dedicated test workflows for Windows and Linux\r\n- Enhanced CI\u002FCD — App building workflows and automated testing\r\n\r\n## Documentation\r\n\r\nNew comprehensive guides for MCP integration, n8n workflows, Jira agent\r\nusage, and app development.\r\n\r\n\r\n\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Febbab2af-7051-4ae8-9064-2a651489d54d","2025-09-30T21:09:42"]