[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-GreatScottyMac--RooFlow":3,"tool-GreatScottyMac--RooFlow":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":76,"owner_company":76,"owner_location":76,"owner_email":76,"owner_twitter":76,"owner_website":76,"owner_url":78,"languages":79,"stars":96,"forks":97,"last_commit_at":98,"license":99,"difficulty_score":23,"env_os":100,"env_gpu":101,"env_ram":101,"env_deps":102,"category_tags":107,"github_topics":76,"view_count":23,"oss_zip_url":76,"oss_zip_packed_at":76,"status":16,"created_at":108,"updated_at":109,"faqs":110,"releases":156},2027,"GreatScottyMac\u002FRooFlow","RooFlow","RooFlow - Enhanced Memory Bank System with ☢️Footgun Power☢️  Next-gen Memory Bank system with five integrated modes and system-level customization. Uses Roo Code's experimental \"Footgun\" feature for deep AI assistant customization while maintaining efficient token usage! ","RooFlow 是一款为 VS Code 用户设计的 AI 辅助开发增强插件，基于 Roo Code 扩展构建，提供五种专注不同任务的智能模式：架构设计、代码生成、调试分析、问答咨询和任务协调。它通过 YAML 格式的系统提示词替代传统 Markdown，显著降低 AI 调用时的 token 消耗，提升响应效率。RooFlow 引入“记忆银行”机制，自动保存并关联项目上下文，让 AI 在多轮对话中持续理解你的代码结构与开发意图，无需重复说明背景。其核心亮点是五种模式协同工作，由“协调者”智能分派任务，兼顾效率与精准度，同时支持实时更新记忆内容。适合频繁使用 AI 编程助手的开发者，尤其是中大型项目中的工程师，希望减少重复提示、提升开发连贯性的人群。虽然功能实验性较强，不属官方支持，但对追求高效、低开销 AI 开发体验的用户而言，是一个值得尝试的轻量级优化方案。","\u003Cdiv align=\"center\">\n\n## ⚠️ Alternate System Prompt Format for Roo Code ⚠️\n\n**RooFlow provides an *experimental* set of modes and system prompts designed as an alternative way to interact with the [Roo Code](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code) VS Code extension.** It uses YAML-based system prompts aiming for improved efficiency and token usage compared to the standard Markdown prompts. RooFlow is NOT officially endorsed or supported by the Roo Code project. Use with caution.\n\n### ✨Now with optional installation for use with the new [Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal)!!✨\n\n\u003Cbr>\n\n# RooFlow \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGreatScottyMac_RooFlow_readme_34d47d9d027a.png\" alt=\"RooFlow Logo\" width=\"300\"\u002F>\n\n**Persistent Project Context and Streamlined AI-Assisted Development**\n\n[![Roo Code](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FVS%20Code-Extension-blue.svg)](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code)\n[![RooFlow](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FView%20on-GitHub-lightgrey.svg)](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow)\n\n\u003C\u002Fdiv>\n\n## 🎯 Overview\n\nRooFlow offers an **experimental alternative** approach to AI-assisted development within the **[Roo Code](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code) VS Code extension**. It utilizes **YAML-based system prompts** for its specialized \"Flow\" modes, potentially offering **improved efficiency and reduced token consumption** compared to standard modes. RooFlow provides **persistent project context** through its Memory Bank system (adapted from the concepts in [roo-code-memory-bank](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Froo-code-memory-bank)), ensuring the AI assistant maintains project understanding across sessions.\n\n### Key Improvements over Roo Code Memory Bank:\n\n*   **Reduced Token Consumption:** Optimized prompts and instructions minimize token usage.\n*   **Five Integrated Modes:** Flow-Architect, Flow-Code, Flow-Debug, Flow-Ask, and Flow-Orchestrator modes work together seamlessly.\n*   **Simplified Setup:**  Easier installation and configuration.\n*   **Streamlined Real-time Updates:**  More efficient and targeted Memory Bank updates.\n*   **Clearer Instructions:**  Improved YAML-based rule files for better readability and maintainability.\n\n### Key Components\n\n```mermaid\n%%{init: {'themeVariables': { 'fontSize': '14px'}}}%%\ngraph LR\n    subgraph Workspace\n        U[User]\n        MB[(Memory Bank)]\n    end\n\n    subgraph VSCode\n        RC[Roo Code Extension]\n    end\n\n    subgraph RooFlow [\"RooFlow (Experimental Alternative)\"]\n        subgraph FlowModes [\"Flow Modes (YAML Prompts)\"]\n            direction TB\n            FO[Flow-Orchestrator]\n            subgraph ExecutionModes [\"Direct\u002FDelegated Modes\"]\n               direction LR\n               FA[Flow-Architect]\n               FC[Flow-Code]\n               FD[Flow-Debug]\n               FK[Flow-Ask]\n            end\n             %% Link Orchestrator to the group it delegates to\n             FO -- Delegates --> ExecutionModes\n        end\n    end\n\n    %% Core Flow\n    U -- Interacts with --> RC\n    RC -- Uses --> RooFlow\n\n    %% User Task Paths\n    U -- Complex Task --> FO\n    U -- Direct Task --> ExecutionModes\n\n\n    %% Memory Bank Interactions\n    FA \u003C--> MB\n    FC \u003C--> MB\n    FD \u003C--> MB\n    FK --> MB\n\n    %% Styling\n    style MB fill:#f9f,stroke:#333,stroke-width:2px,color:#000\n    style FO fill:#ccf,stroke:#333,stroke-width:2px,color:#000\n```\n\n**Memory Bank Access Legend:**\n*   `\u003C-->` (Read\u002FWrite): Flow-Architect, Flow-Code, Flow-Debug\n*   `-->` (Read-Only): Flow-Ask\n\n\n- 🧠 **Memory Bank**: Persistent storage for project knowledge (automatically managed).\n- 💻 **System Prompts**: YAML-based core instructions for each mode (`.roo\u002Fsystem-prompt-[mode]`).\n- 🔧 **VS Code Integration**: Seamless development experience within VS Code.\n- ⚡ **Real-time Updates**:  Automatic Memory Bank updates triggered by significant events.\n\n## 🚀 Quick Start\n\n   ### Installation\n\n   1.  **Install Roo Code Extension:** Ensure you have the [Roo Code extension](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code) installed in VS Code.\n   2.  **Prerequisites:**\n       *   **Git:** The installation script requires `git` to be installed and accessible in your system's PATH. Download Git from [https:\u002F\u002Fgit-scm.com\u002Fdownloads](https:\u002F\u002Fgit-scm.com\u002Fdownloads).\n       *   **Python 3:** The script now uses Python directly. Ensure Python 3 is installed and accessible as `python3` (Linux\u002FmacOS) or `python` (Windows) in your PATH. Download Python from [https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F).\n       *   **PyYAML:** The Python script requires the PyYAML library. Install it using pip:\n           ```bash\n           pip install pyyaml\n           # or potentially pip3 install pyyaml\n           ```\n   3.  **Open your terminal** and navigate (`cd`) to your project's **root directory**.\n   4.  **Run the appropriate command** for your operating system directly:\n       *   **Windows (Command Prompt or PowerShell):**\n           1. Download the script\n\n              **For Use With File Based Memory Bank:**\n              ```cmd\n              curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.cmd\n              ```\n              **For Use With [Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal):**\n              ```cmd\n              curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.cmd\n              ```\n           2. Execute the downloaded script (⚠️If you have MCP servers connected, see [Importing Connected MCP Server Tools](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow?tab=readme-ov-file#importing-connected-mcp-server-tools-optional) prior to installation):\n              ```bash\n              .\u002Finstall_rooflow.cmd\n              ```\n              or \n              ```bash\n              .\u002Finstall_rooflow_conport.cmd\n              ```\n       *   **Linux \u002F macOS (bash\u002Fzsh):**\n           1. Download the script\n\n              **For Use With File Based Memory Bank:**\n              ```cmd\n              curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.sh\n              ```\n              **For Use With [Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal):**\n              ```cmd\n              curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.sh\n              ```\n           2. Make the script executable:\n              ```bash\n              chmod +x install_rooflow.sh\n              ```\n              or  \n              ```bash\n              chmod +x install_rooflow_conport.sh\n              ```\n           3. Execute the downloaded script (⚠️If you have MCP servers connected, see [Importing Connected MCP Server Tools](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow?tab=readme-ov-file#importing-connected-mcp-server-tools-optional) prior to installation):\n              ```bash\n              .\u002Finstall_rooflow.sh\n              ```\n              or \n              ```bash\n              .\u002Finstall_rooflow_conport.sh\n              ```\n   5.  The command downloads and executes the script, which will check for prerequisites (`git`, `python3`\u002F`python`, `pyyaml`), clone the repository, copy necessary files (including `generate_mcp_yaml.py`), clean up the clone, and run the Python script to process templates. Follow any on-screen prompts or error messages.\n   6.  **Note:** The installer script (`install_rooflow.*`) is designed to remain after execution for potential re-runs (e.g., for updates).\n\n   7.  **Verify Installation:** After the script runs successfully:\n       *   Check that the `.roo\u002F` directory, along with the `.roomodes`file exist in your project root.\n       *   Optionally, inspect the `.roo\u002Fsystem-prompt-*` files to ensure placeholders like `WORKSPACE_PLACEHOLDER` have been replaced with your actual system paths.\n   8.  **Coexistence with Standard Modes:** The RooFlow installation adds the `Flow-*` modes alongside the standard Roo Code modes (Architect, Code, Debug, Ask). You can switch between Flow modes and standard modes at any time using the Roo Code interface. If you encounter issues with a Flow mode, you can easily switch back to its standard counterpart.\n\n   ### Importing Connected MCP Server Tools (Optional)\n\n   If you use [MCP (Model Context Protocol)](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fspecification) servers with Roo Code, RooFlow can automatically inject the details of your connected servers (their tools, resources, etc.) into the custom Flow mode prompts during installation. This ensures the Flow modes are aware of the same tools as the standard Roo Code modes.\n\n   **To enable this during initial installation:**\n\n   1.  **Get Full System Prompt:** *Before* running the RooFlow installer, you need the complete system prompt text that Roo Code generates dynamically *after* it connects to your MCP servers.\n       *   Navigate to Extensions -> Roo Code -> Prompts (as shown in illustration below).\n       \u003Cbr\u002F>\n       \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGreatScottyMac_RooFlow_readme_e9b6304e3af2.png\" alt=\"Create Mode View\" width=\"200\"\u002F>\n\n       *   Select any *standard* Roo Code mode (e.g., \"Architect\", \"Code\", \"Ask\", \"Debug\").\n       *   Click the \"Copy system prompt to clipboard\" button (as shown in illustration above).\n       *   This text includes the dynamically discovered information about your *currently connected* MCP servers.\n   2.  **Save as `system_prompt.md`:** Save this copied text into a new file named exactly `system_prompt.md` directly in your project's **root directory**.\n   3.  **Run Installer:** Now, run the `install_rooflow.sh` or `install_rooflow.cmd` script as described in the main Installation section above.\n       *   The installer copies the Python script `generate_mcp_yaml.py`.\n       *   It then runs this Python script (using `python3` or `python`). The script automatically looks for `system_prompt.md` in the root directory.\n       *   If found, the Python script parses the MCP server details from it, converts them to YAML, and injects them into the appropriate placeholder (`# [CONNECTED_MCP_SERVERS]`) within the `.roo\u002Fsystem-prompt-flow-*` files while also performing other variable substitutions.\n       *   If `system_prompt.md` is not found, this injection step is simply skipped.\n\n   **Updating MCP Server Information Later:**\n\n   If you add, remove, or modify your MCP server configurations after the initial RooFlow installation, you'll need to update the Flow mode prompts:\n\n   1.  **Get Updated Prompt:** Ensure your desired MCP servers are connected in Roo Code. Go back to Roo Code settings and copy the *latest* full system prompt text (which includes the updated MCP details) from a standard mode, just like in the initial setup.\n   2.  **Save Updated `system_prompt.md`:** Save this new text into `system_prompt.md` in your project root, overwriting the previous version.\n   3.  **Re-run Installer:** Execute the `install_rooflow.sh` or `install_rooflow.cmd` script again from your project root.\n       *   This script will copy the latest base configuration (`.roo\u002F` files, `.roomodes`, `generate_mcp_yaml.py`) from the repository *and* run the `generate_mcp_yaml.py` Python script again.\n       *   The Python script will read your updated `system_prompt.md` and inject the new MCP server details into the freshly copied `.roo\u002Fsystem-prompt-flow-*` files (along with performing basic variable substitutions).\n\n   **⚠️ Important Warning about Updates:** Re-running the installer script is the simplest way to update MCP info, but be aware that it **overwrites** the `.roo\u002F` directory and the `.roomodes` file with fresh copies from the repository. Any manual customizations you might have made directly to the `system-prompt-flow-*` files *after* the initial installation will be **lost**. The Python script (`generate_mcp_yaml.py`) remains in the root if you wish to build a more granular update process manually.\n\n\n   ### 3. Using RooFlow\n\n   1.  **Start a Chat:** Open a new Roo Code chat in your project.\n   2.  **Select a Mode:** Choose the appropriate mode for your task:\n       *   **🌊Flow Code💻:** For writing, modifying, and documenting code.\n       *   **🌊Flow Architect🏗️:** For system design, project structure, and Memory Bank initialization.\n       *   **🌊Flow Ask❓:** For questions, code analysis, and explanations.\n       *   **🌊Flow Debug🪲:** For troubleshooting errors and debugging code.\n       *   **🌊Flow Orchestrator🪃:** For complex tasks requiring breakdown and delegation to other modes (see [Boomerang Tasks](#boomerang-tasks-orchestrate-complex-workflows)).\n   3.  **Interact with Roo:** Give Roo instructions and ask questions. Roo will use the Memory Bank (if initialized) and mode capabilities to assist.\n   4.  **Memory Bank Initialization:**  If you start a chat in a project *without* a `memory-bank\u002F` directory, Roo will suggest switching to Flow-Architect mode and guide you through the initialization process.\n   5. **\"Update Memory Bank\" Command:** At any time, you can type \"Update Memory Bank\" or \"UMB\" to force a synchronization of the chat session's information into the Memory Bank. This is useful for ensuring continuity across sessions or before switching modes.\n\n   \u003Cbr\u002F>\n\n   ### 4. Updating RooFlow\n\n   #### Simply run the install script and it will overwrite your existing .roo\u002F directory and .roomodes file. If you have your own additional custom modes in .roomodes, rename that file before running the update and then replace the contents of the newly installed .roomodes with the content from your custom file.\n\n\u003Cbr\u002F>\n\n## 📚 Memory Bank Structure\n\nThe Memory Bank is a directory named `memory-bank` located in your project's root. It contains several Markdown files that store different aspects of your project's knowledge:\n\n| File                 | Purpose                                                                                                                               |\n| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |\n| `activeContext.md`   | Tracks the current session's context: recent changes, current goals, and open questions\u002Fissues.                                       |\n| `decisionLog.md`     | Records architectural and implementation decisions, including the context, decision, rationale, and implementation details.        |\n| `productContext.md`  | Provides a high-level overview of the project, including its goals, features, and overall architecture.                             |\n| `progress.md`        | Tracks the progress of the project, including completed work, current tasks, and next steps.  Uses a task list format.               |\n| `systemPatterns.md` | (Optional) Documents recurring patterns and standards used in the project (coding patterns, architectural patterns, testing patterns). |\n\nRooFlow automatically manages these files. You generally don't need to edit them directly, although you can review them to understand the AI's knowledge.\n\n## ✨ Features\n\n### 🧠 Persistent Context\n\nRooFlow remembers project details across sessions, maintaining a consistent understanding of your codebase, design decisions, and progress.\n\n### ⚡ Real-time Updates\n\nThe Memory Bank is updated automatically based on significant events within each mode, ensuring that the context is always up-to-date.\n\n### 🤝 Mode Collaboration\n\nThe five modes (Flow-Architect, Flow-Code, Flow-Debug, Flow-Ask, and Flow-Orchestrator) are designed to work together seamlessly. Flow-Orchestrator can delegate tasks to the other modes, and they can switch between each other as needed, sharing information through the Memory Bank.\n\n### ⬇️ Reduced Token Consumption\n\nRooFlow is designed to use fewer tokens than previous systems, making it more efficient and cost-effective.\n\n## 📝 UMB Command\nThe command \"Update Memory Bank\" or \"UMB\" can be given at any time to update the memory bank with information from the current chat session.\n\n## Contributing\n\nContributions to RooFlow are welcome! Please see the [CONTRIBUTING.md](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fblob\u002Fmain\u002FCONTRIBUTING.md) file (you'll need to create this) for guidelines.\n\n## License\n  [Apache 2.0](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fblob\u002Fmain\u002FLICENSE>)\n","\u003Cdiv align=\"center\">\n\n## ⚠️ Roo Code 的备用系统提示格式 ⚠️\n\n**RooFlow 提供了一组*实验性*模式和系统提示，旨在作为与 [Roo Code](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code) VS Code 扩展交互的替代方式。** 它采用基于 YAML 的系统提示，与标准 Markdown 提示相比，可提高效率并减少令牌使用量。RooFlow 并未获得 Roo Code 项目的官方认可或支持。请谨慎使用。\n\n### ✨现支持可选安装，以配合全新的 [Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal) 使用！！✨\n\n\u003Cbr>\n\n# RooFlow \n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGreatScottyMac_RooFlow_readme_34d47d9d027a.png\" alt=\"RooFlow Logo\" width=\"300\"\u002F>\n\n**持久化项目上下文与精简的 AI 辅助开发**\n\n[![Roo Code](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FVS%20Code-Extension-blue.svg)](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code)\n[![RooFlow](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FView%20on-GitHub-lightgrey.svg)](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow)\n\n\u003C\u002Fdiv>\n\n## 🎯 概览\n\nRooFlow 为 **[Roo Code](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code) VS Code 扩展** 中的 AI 辅助开发提供了一种**实验性的替代方案**。它采用**基于 YAML 的系统提示**来实现其特有的“Flow”模式，相较于标准模式，可能带来**更高的效率和更低的令牌消耗**。RooFlow 通过其 Memory Bank 系统（借鉴了 [roo-code-memory-bank](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Froo-code-memory-bank) 中的概念）提供了**持久化的项目上下文**，确保 AI 助手在不同会话中都能保持对项目的理解。\n\n### 相较于 Roo Code Memory Bank 的关键改进：\n\n*   **降低令牌消耗：** 优化后的提示和指令最大限度地减少了令牌使用。\n*   **五种集成模式：** Flow-Architect、Flow-Code、Flow-Debug、Flow-Ask 和 Flow-Orchestrator 模式无缝协作。\n*   **简化设置：** 更加简便的安装与配置。\n*   **精简实时更新：** Memory Bank 更新更高效且更具针对性。\n*   **更清晰的说明：** 改进的 YAML 规则文件，提升可读性和可维护性。\n\n### 关键组件\n\n```mermaid\n%%{init: {'themeVariables': { 'fontSize': '14px'}}}%%\ngraph LR\n    subgraph Workspace\n        U[用户]\n        MB[(Memory Bank)]\n    end\n\n    subgraph VSCode\n        RC[Roo Code 扩展]\n    end\n\n    subgraph RooFlow [\"RooFlow（实验性替代方案）\"]\n        subgraph FlowModes [\"Flow 模式（YAML 提示）\"]\n            direction TB\n            FO[Flow-Orchestrator]\n            subgraph ExecutionModes [\"直接\u002F委托模式\"]\n               direction LR\n               FA[Flow-Architect]\n               FC[Flow-Code]\n               FD[Flow-Debug]\n               FK[Flow-Ask]\n            end\n             %% 将 Orchestrator 链接到它所委托的组\n             FO -- 委托 --> ExecutionModes\n        end\n    end\n\n    %% 核心流程\n    U -- 与 --> RC 交互\n    RC -- 使用 --> RooFlow\n\n    %% 用户任务路径\n    U -- 复杂任务 --> FO\n    U -- 直接任务 --> ExecutionModes\n\n\n    %% Memory Bank 交互\n    FA \u003C--> MB\n    FC \u003C--> MB\n    FD \u003C--> MB\n    FK --> MB\n\n    %% 样式\n    style MB fill:#f9f,stroke:#333,stroke-width:2px,color:#000\n    style FO fill:#ccf,stroke:#333,stroke-width:2px,color:#000\n```\n\n**Memory Bank 访问标记：**\n*   `\u003C-->`（读\u002F写）：Flow-Architect、Flow-Code、Flow-Debug\n*   `-->`（只读）：Flow-Ask\n\n\n- 🧠 **Memory Bank**：用于存储项目知识的持久化存储（自动管理）。\n- 💻 **系统提示**：每个模式的核心指令均基于 YAML（位于 `.roo\u002Fsystem-prompt-[mode]`）。\n- 🔧 **VS Code 集成**：在 VS Code 中实现无缝开发体验。\n- ⚡ **实时更新**：由重要事件触发的 Memory Bank 自动更新。\n\n## 🚀 快速入门\n\n### 安装\n\n1. **安装 Roo Code 扩展：** 确保您已在 VS Code 中安装了 [Roo Code 扩展](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code)。\n2. **先决条件：**\n   *   **Git：** 安装脚本需要 `git` 已安装并可在系统的 PATH 中访问。请从 [https:\u002F\u002Fgit-scm.com\u002Fdownloads](https:\u002F\u002Fgit-scm.com\u002Fdownloads) 下载 Git。\n   *   **Python 3：** 脚本现在直接使用 Python。确保 Python 3 已安装，并可在 PATH 中以 `python3`（Linux\u002FmacOS）或 `python`（Windows）访问。请从 [https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F) 下载 Python。\n   *   **PyYAML：** Python 脚本需要 PyYAML 库。使用 pip 安装：\n     ```bash\n     pip install pyyaml\n     # 或者可能需要 pip3 install pyyaml\n     ```\n3. **打开您的终端**，并导航（cd）到项目的**根目录**。\n4. **直接运行适用于您操作系统的相应命令：**\n   *   **Windows（命令提示符或 PowerShell）：**\n     1. 下载脚本\n\n        **用于基于文件的 Memory Bank：**\n        ```cmd\n        curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.cmd\n        ```\n        **用于 [Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal)：**\n        ```cmd\n        curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.cmd\n        ```\n     2. 执行下载的脚本（⚠️如果您已连接 MCP 服务器，请先参阅 [导入已连接的 MCP 服务器工具](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow?tab=readme-ov-file#importing-connected-mcp-server-tools-optional) 再进行安装）：\n        ```bash\n        .\u002Finstall_rooflow.cmd\n        ```\n        或\n        ```bash\n        .\u002Finstall_rooflow_conport.cmd\n        ```\n   *   **Linux \u002F macOS（bash\u002Fzsh）：**\n     1. 下载脚本\n\n**适用于基于文件的内存库：**\n```cmd\ncurl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.sh\n```\n\n**适用于[Context Portal MCP](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002Fcontext-portal)：**\n```cmd\ncurl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.sh\n```\n\n2. 使脚本可执行：\n```bash\nchmod +x install_rooflow.sh\n```\n或\n```bash\nchmod +x install_rooflow_conport.sh\n```\n\n3. 执行下载的脚本（⚠️如果您已连接MCP服务器，请在安装前参阅[导入已连接的MCP服务器工具](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow?tab=readme-ov-file#importing-connected-mcp-server-tools-optional)）：\n```bash\n.\u002Finstall_rooflow.sh\n```\n或\n```bash\n.\u002Finstall_rooflow_conport.sh\n```\n\n5. 该命令会下载并执行脚本，脚本将检查先决条件（`git`、`python3`\u002F`python`、`pyyaml`），克隆仓库，复制必要文件（包括`generate_mcp_yaml.py`），清理克隆目录，并运行Python脚本处理模板。请根据屏幕提示或错误信息进行操作。\n\n6. **注意：** 安装脚本（`install_rooflow.*`）在执行后会保留下来，以便后续重新运行（例如用于更新）。\n\n7. **验证安装：** 脚本成功运行后：\n* 检查项目根目录中是否存在`.roo\u002F`目录以及`.roomodes`文件。\n* 可选地，检查`.roo\u002Fsystem-prompt-*`文件，确保诸如`WORKSPACE_PLACEHOLDER`之类的占位符已被替换为您的实际系统路径。\n\n8. **与标准模式共存：** RooFlow安装会将`Flow-*`模式与标准Roo Code模式（Architect、Code、Debug、Ask）并列添加。您可随时通过Roo Code界面在Flow模式和标准模式之间切换。如果遇到Flow模式的问题，您可以轻松切换回其标准对应模式。\n\n### 导入已连接的MCP服务器工具（可选）\n\n如果您在Roo Code中使用[MCP（模型上下文协议）](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fspecification)服务器，RooFlow可在安装时自动将已连接服务器的详细信息（包括其工具、资源等）注入自定义Flow模式提示中。这确保了Flow模式能够识别与标准Roo Code模式相同的工具。\n\n**要在初始安装时启用此功能：**\n\n1. **获取完整系统提示：** 在运行RooFlow安装程序之前，您需要Roo Code动态生成的完整系统提示文本——该文本是在Roo Code连接到您的MCP服务器后生成的。\n* 导航至“扩展”->“Roo Code”->“提示”（如图所示）。\n\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGreatScottyMac_RooFlow_readme_e9b6304e3af2.png\" alt=\"创建模式视图\" width=\"200\"\u002F>\n\n* 选择任意一个*标准*Roo Code模式（例如，“Architect”、“Code”、“Ask”、“Debug”）。\n* 点击“复制系统提示到剪贴板”按钮（如上图所示）。\n* 此文本包含您当前已连接的MCP服务器的动态发现信息。\n\n2. **保存为`system_prompt.md`：** 将复制的文本保存到一个名为`system_prompt.md`的新文件中，直接放在项目的**根目录**下。\n\n3. **运行安装程序：** 现在，按照上述主要安装部分所述，运行`install_rooflow.sh`或`install_rooflow.cmd`脚本。\n* 安装程序会复制Python脚本`generate_mcp_yaml.py`。\n* 随后运行该Python脚本（使用`python3`或`python`）。脚本会自动在根目录中查找`system_prompt.md`。\n* 如果找到，Python脚本会解析其中的MCP服务器详细信息，将其转换为YAML格式，并注入到`.roo\u002Fsystem-prompt-flow-*`文件中的相应占位符（`# [CONNECTED_MCP_SERVERS]`）中，同时执行其他变量替换。\n* 如果未找到`system_prompt.md`，则跳过此注入步骤。\n\n**稍后更新MCP服务器信息：**\n\n如果您在RooFlow初始安装后添加、删除或修改MCP服务器配置，您需要更新Flow模式提示：\n\n1. **获取更新后的提示：** 确保所需的MCP服务器已在Roo Code中连接。返回Roo Code设置，从一个标准模式中复制最新的完整系统提示文本（包含更新后的MCP详细信息），就像初始设置一样。\n2. **保存更新后的`system_prompt.md`：** 将新文本保存到项目根目录下的`system_prompt.md`中，覆盖之前的版本。\n3. **重新运行安装程序：** 从项目根目录再次执行`install_rooflow.sh`或`install_rooflow.cmd`脚本。\n* 该脚本会从仓库中复制最新的基础配置（`.roo\u002F`文件、`.roomodes`、`generate_mcp_yaml.py`），并再次运行`generate_mcp_yaml.py` Python脚本。\n* Python脚本会读取您更新后的`system_prompt.md`，并将新的MCP服务器详细信息注入到新复制的`.roo\u002Fsystem-prompt-flow-*`文件中（同时执行基本变量替换）。\n\n**⚠️ 关于更新的重要提示：** 重新运行安装程序是更新MCP信息最简单的方法，但请注意，这会**覆盖**`.roo\u002F`目录和`.roomodes`文件，用仓库中的最新副本替换它们。如果您在初始安装后对`.roo\u002Fsystem-prompt-flow-*`文件进行了手动自定义，这些自定义内容将会**丢失**。如果您希望构建更精细的更新流程，Python脚本（`generate_mcp_yaml.py`）仍保留在根目录中。\n\n### 3. 使用RooFlow\n\n1. **开始聊天**：在您的项目中打开一个新的Roo Code聊天。\n   2. **选择模式**：为您的任务选择合适的模式：\n       *   **🌊Flow代码💻**：用于编写、修改和记录代码。\n       *   **🌊Flow架构师🏗️**：用于系统设计、项目结构以及Memory Bank的初始化。\n       *   **🌊Flow提问❓**：用于提问、代码分析和解释。\n       *   **🌊Flow调试🪲**：用于排查错误和调试代码。\n       *   **🌊Flow协调员🪃**：用于需要分解并委托给其他模式的复杂任务（参见[回旋镖任务](#boomerang-tasks-orchestrate-complex-workflows)）。\n   3. **与Roo互动**：向Roo发出指令并提出问题。Roo将利用Memory Bank（如果已初始化）和模式功能来提供帮助。\n   4. **Memory Bank初始化**：如果您在没有`memory-bank\u002F`目录的项目中开始聊天，Roo会建议切换到Flow-架构师模式，并引导您完成初始化过程。\n   5. **“更新Memory Bank”命令**：您可以随时输入“更新Memory Bank”或“UMB”，以强制将聊天会话的信息同步到Memory Bank。这在确保跨会话连续性或切换模式前非常有用。\n\n   \u003Cbr\u002F>\n\n   ### 4. 更新RooFlow\n\n   #### 只需运行安装脚本，它就会覆盖您现有的`.roo\u002F`目录和`.roomodes`文件。如果您在`.roomodes`中有自己的额外自定义模式，请在运行更新前重命名该文件，然后用您自定义文件的内容替换新安装的`.roomodes`文件内容。\n\n\u003Cbr\u002F>\n\n\n\n## 📚 Memory Bank结构\n\nMemory Bank是一个名为`memory-bank`的目录，位于您的项目根目录下。它包含若干Markdown文件，用于存储项目知识的不同方面：\n\n| 文件                 | 用途                                                                                                                               |\n| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |\n| `activeContext.md`   | 跟踪当前会话的上下文：最近的变更、当前目标以及开放的问题\u002F议题。                                       |\n| `decisionLog.md`     | 记录架构和实现决策，包括上下文、决策内容、理由及实施细节。                                           |\n| `productContext.md`  | 提供项目的高层次概览，包括其目标、功能和整体架构。                                                     |\n| `progress.md`        | 跟踪项目的进展，包括已完成的工作、当前任务和下一步计划。采用任务列表格式。                             |\n| `systemPatterns.md`  | （可选）记录项目中使用的常见模式和标准（编码模式、架构模式、测试模式）。                               |\n\nRooFlow会自动管理这些文件。通常您无需直接编辑它们，不过可以查看这些文件以了解AI的知识。\n\n## ✨ 功能\n\n### 🧠 持久化上下文\n\nRooFlow会在不同会话之间记住项目细节，保持对您的代码库、设计决策和进度的一致理解。\n\n### ⚡ 实时更新\n\nMemory Bank会根据每个模式中的重要事件自动更新，确保上下文始终是最新的。\n\n### 🤝 模式协作\n\n五个模式（Flow-架构师、Flow-代码、Flow-调试、Flow-提问和Flow-协调员）旨在无缝协作。Flow-协调员可以将任务委派给其他模式，各模式也可根据需要相互切换，并通过Memory Bank共享信息。\n\n### ⬇️ 降低Token消耗\n\nRooFlow的设计旨在比以前的系统使用更少的Token，从而更加高效且经济实惠。\n\n## 📝 UMB命令\n随时可以输入“更新Memory Bank”或“UMB”命令，以将当前聊天会话的信息更新到Memory Bank。\n\n## 贡献\n\n欢迎为RooFlow做出贡献！请参阅[CONTRIBUTING.md](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fblob\u002Fmain\u002FCONTRIBUTING.md)文件（您需要创建此文件）以获取指南。\n\n## 许可证\n[Apache 2.0](https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fblob\u002Fmain\u002FLICENSE>)","# RooFlow 快速上手指南\n\n## 环境准备\n\n- **VS Code**：已安装 [Roo Code 扩展](https:\u002F\u002Fgithub.com\u002FRooVetGit\u002FRoo-Code)\n- **Git**：确保已安装并加入系统 PATH  \n  [下载地址](https:\u002F\u002Fgit-scm.com\u002Fdownloads)（国内推荐：[清华镜像](https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fgit\u002F)）\n- **Python 3**：确保可执行命令为 `python3`（Linux\u002FmacOS）或 `python`（Windows）  \n  [下载地址](https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F)（国内推荐：[清华源](https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\u002F)）\n- **PyYAML**：安装依赖库  \n  ```bash\n  pip install pyyaml\n  # 或使用国内源加速\n  pip install pyyaml -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n  ```\n\n## 安装步骤\n\n1. 打开终端，进入你的项目**根目录**。\n\n2. 根据使用场景选择安装脚本：\n\n   **使用本地 Memory Bank（推荐默认）**  \n   ```bash\n   # Linux\u002FmacOS\n   curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.sh\n   chmod +x install_rooflow.sh\n   .\u002Finstall_rooflow.sh\n\n   # Windows（PowerShell 或 CMD）\n   curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow.cmd\n   .\u002Finstall_rooflow.cmd\n   ```\n\n   **使用 Context Portal MCP（可选）**  \n   ```bash\n   # Linux\u002FmacOS\n   curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.sh\n   chmod +x install_rooflow_conport.sh\n   .\u002Finstall_rooflow_conport.sh\n\n   # Windows\n   curl -O https:\u002F\u002Fraw.githubusercontent.com\u002FGreatScottyMac\u002FRooFlow\u002Fmain\u002Fconfig\u002Finstall_rooflow_conport.cmd\n   .\u002Finstall_rooflow_conport.cmd\n   ```\n\n3. 安装完成后，检查项目根目录是否生成 `.roo\u002F` 文件夹及 `.roomodes` 文件。\n\n## 基本使用\n\n1. 在 VS Code 中打开你的项目。\n2. 打开 Roo Code 面板，选择任意 **Flow 模式**：\n   - `Flow-Architect`：架构设计\n   - `Flow-Code`：代码生成\n   - `Flow-Debug`：调试分析\n   - `Flow-Ask`：问答咨询\n   - `Flow-Orchestrator`：任务编排（推荐复杂任务）\n\n3. 输入你的开发请求，例如：\n   ```\n   请为这个项目添加一个用户登录模块，使用 Python + Flask\n   ```\n\n4. RooFlow 会自动利用项目上下文（Memory Bank）进行响应，无需重复提供背景信息。\n\n> ✅ 所有 Flow 模式与原 Roo Code 模式共存，可随时切换。","一位全栈开发者正在维护一个包含 15 个微服务的遗留系统，需在两周内为新需求添加 API 接口并修复三个历史 Bug。他每天要反复向 AI 助手询问上下文：哪个服务调用了哪个数据库？这个字段在哪个模块被修改过？上次修复类似 Bug 的提交是哪次？\n\n### 没有 RooFlow 时\n- 每次重启 VS Code，AI 都“忘记”项目结构，必须重新粘贴 README 和架构图作为上下文，浪费大量时间。\n- 使用标准 Markdown 提示时，单次请求消耗 2000+ token，频繁触发限流，响应延迟超过 10 秒。\n- 在调试时，AI 无法关联不同服务间的调用链，常给出错误的依赖推测，导致误改代码。\n- 项目中多个模块的命名规则不一致，AI 无法自动识别“UserSvc”和“userService”是同一服务。\n- 每次切换任务（如从开发转为调试），需手动重设 AI 角色，流程零散、效率低下。\n\n### 使用 RooFlow 后\n- Memory Bank 自动记录所有服务名、接口定义和关键变更历史，即使关闭 VS Code，重启后 AI 仍能精准回忆上下文。\n- YAML 格式提示词将单次请求 token 消耗降低至 800 以内，响应速度提升 60%，不再因限流中断工作流。\n- Flow-Debug 模式能自动关联跨服务调用链，结合 Memory Bank 中的日志路径，准确指出问题发生在 OrderService 的缓存失效环节。\n- Flow-Architect 模式自动标准化命名识别，将“UserSvc”“userService”统一映射为“用户服务”，避免混淆。\n- 通过 Flow-Orchestrator 一键切换模式，开发、调试、提问无缝衔接，无需重复配置，任务切换效率提升近 70%。\n\nRooFlow 让 AI 从“每次都要重新教”的助手，变成了真正懂你项目的“第二大脑”。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FGreatScottyMac_RooFlow_34d47d9d.png","GreatScottyMac",null,"https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002FGreatScottyMac_0560f2c3.png","https:\u002F\u002Fgithub.com\u002FGreatScottyMac",[80,84,88,92],{"name":81,"color":82,"percentage":83},"Python","#3572A5",30,{"name":85,"color":86,"percentage":87},"Shell","#89e051",29,{"name":89,"color":90,"percentage":91},"Batchfile","#C1F12E",28.3,{"name":93,"color":94,"percentage":95},"PowerShell","#012456",12.7,1117,122,"2026-04-05T00:54:07","Apache-2.0","Linux, macOS, Windows","未说明",{"notes":103,"python":104,"dependencies":105},"需要安装 Git 和 Python 3，并确保 pip 可用；首次安装会自动克隆仓库并生成配置文件；若使用 Context Portal MCP，需提前复制 Roo Code 的系统提示到 system_prompt.md 文件中；重新运行安装脚本会覆盖 .roo\u002F 目录下的自定义配置。","3.8+",[106],"PyYAML",[53,15,13],"2026-03-27T02:49:30.150509","2026-04-06T06:53:14.523131",[111,116,121,126,131,135,140,144,148,152],{"id":112,"question_zh":113,"answer_zh":114,"source_url":115},9203,"为什么 RooFlow 无法使用 MCP 服务器（如 MySQL）？","RooFlow 的系统提示中默认未包含 MCP 工具的描述，导致 LLM 无法识别或调用 MCP 服务。解决方法是手动将 MCP 工具的定义添加到 system-prompt-[mode].txt 文件中，例如添加类似 \u003Cread_file>、\u003Cfetch_instructions> 等工具的 XML 格式说明，确保 LLM 知道可用的工具及其调用方式。维护者已为此更新了系统提示文件。","https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fissues\u002F7",{"id":117,"question_zh":118,"answer_zh":119,"source_url":120},9204,"如何解决 RooFlow 工具调用只显示不执行的问题？","该问题通常由系统提示过时或格式错误导致。请运行最新的 insert-variables.cmd 脚本更新系统提示，并确保工具调用使用正确的 XML 标签格式（如 \u003Cread_file> 而非其他变体）。若问题仍存在，尝试重命名系统提示文件、重启 VSCode、恢复原文件，以清除缓存的错误提示。","https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fissues\u002F55",{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},9205,"安装 RooFlow 时提示 insert-variables.cmd 文件不存在，如何解决？","该问题因安装脚本路径错误导致。请确保从最新版本重新克隆仓库，因为旧版本的 install_rooflow.cmd 已被修复，不再错误复制 install_rooflow.sh 文件。若仍缺失，手动运行 insert-variables.cmd 脚本（位于 config\u002F 目录），确保其在安装过程中正确替换 HOME_PLACEHOLDER 和 WORKSPACE_PLACEHOLDER 为实际路径。","https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fissues\u002F30",{"id":127,"question_zh":128,"answer_zh":129,"source_url":130},9206,"RooFlow 是否支持 Codebase Indexing（代码库索引）工具？","支持。需在系统提示中正确声明 codebase_search 工具的 XML 格式。维护者已通过 PR #77 添加支持，确保在 system-prompt 中包含类似 \u003Ccodebase_search> 的标签，并使用官方文档中的参数格式（如路径、查询关键词），以确保 LLM 能正确解析和调用该工具。","https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fissues\u002F68",{"id":132,"question_zh":133,"answer_zh":134,"source_url":125},9207,"Windows 用户在 YAML 配置文件中遇到路径分隔符问题，是否需要转义？","不需要转义。.roo 目录中的 home_directory 和 working_directory 字段使用的是占位符（如 HOME_PLACEHOLDER），并非真实路径。安装脚本 insert-variables.cmd 会在运行时自动替换为当前系统的正确路径（如 C:\\Users\\xxx），无需手动修改或转义反斜杠。",{"id":136,"question_zh":137,"answer_zh":138,"source_url":139},9208,"RooFlow 与 MCP Memory Bank 的主要区别是什么？","MCP Memory Bank 主要用于跟踪人员、组织、事件等通用实体上下文，而 RooFlow 专注于为 AI 助手提供项目级上下文（如代码结构、任务目标），并显著减少 token 消耗。RooFlow 更适合开发环境中的代码辅助，Memory Bank 更适合通用知识管理。","https:\u002F\u002Fgithub.com\u002FGreatScottyMac\u002FRooFlow\u002Fissues\u002F4",{"id":141,"question_zh":142,"answer_zh":143,"source_url":120},9209,"更新 RooFlow 后系统提示失效，如何恢复工具调用功能？","每次更新 Roo Code 或 RooFlow 后，必须重新运行 insert-variables.cmd 脚本，以同步最新的系统提示模板。若工具仍无法调用，检查 system-prompt-[mode].txt 是否包含正确的工具定义（如 \u003Cread_file>、\u003Ccodebase_search>），并确保 XML 标签名称与官方文档完全一致。",{"id":145,"question_zh":146,"answer_zh":147,"source_url":115},9210,"为什么使用 Boomerang 模式时，AI 总是拒绝调用 MCP 工具？","AI 拒绝调用是因为系统提示中未明确声明 MCP 工具的存在。请编辑 system-prompt-boomerang.txt，添加类似 '可用工具包括：read_file, codebase_search, fetch_instructions' 的说明，并确保每个工具都以标准 XML 格式定义（如 \u003Cread_file>\u003Cpath>...\u003C\u002Fpath>\u003C\u002Fread_file>），以增强 LLM 对工具的认知。",{"id":149,"question_zh":150,"answer_zh":151,"source_url":130},9211,"如何确认 MCP 工具是否被正确加载到 RooFlow 中？","打开 .roo\u002Fsystem-prompt-[mode].txt 文件，搜索工具名称（如 read_file、codebase_search）。若不存在，说明未正确注入。运行 insert-variables.cmd 脚本后，检查文件是否包含完整的 XML 工具定义块。若仍缺失，手动从官方文档复制工具定义并粘贴到对应文件中。",{"id":153,"question_zh":154,"answer_zh":155,"source_url":120},9212,"使用 AWS Claude 3.5 Sonnet 时 RooFlow 表现不稳定，如何优化？","确保系统提示中工具定义格式严格遵循 XML 规范（无多余空格、标签闭合正确）。避免在提示中使用模糊指令（如 '使用 exa search tool'），应明确使用标准工具名（如 \u003Ccodebase_search>）。同时，更新到 RooFlow 0.4.0+ 版本，该版本已优化提示结构以适配 Claude 3.5 的推理模式。",[157,162],{"id":158,"version":159,"summary_zh":160,"released_at":161},106620,"v0.5.6","v0.5.6\r\n\r\n- Created more detailed support for codebase_search in the RooFlow system prompts.\r\n- Removed the optional CLI command suggestion from the attempt_completion tool parameters.\r\n- Removed the execute_command tool from Flow-Architect. \r\n\r\n\r\nv0.5.5\r\n\r\nAdded  system prompt instructions to support codebase_search. Thanks @edxeth !!\r\n\r\n\r\nv0.5.4\r\n\r\nRemoved the Roo Code modes from the flow-orchestrator system prompt. \r\n\r\n\r\nv0.5.3\r\n\r\n- Removed references to unavailable tools in the Flow-Orchestrator system prompt.\r\n- Added Flow-Orchestrator to the Modes list in the RooFlow system prompts.\r\n- Gave Flow-Orchestrator read-only access to memory-bank\u002F and full access to context_portal\u002F if used.\r\n\r\n\u003Cbr>\r\n\r\nv0.5.2\r\n\r\n- Improved handling of apply_diff errors.\r\n- Added Markdown Formatting Rules section. \r\n\r\n\u003Cbr>\r\n\r\nv0.5.1\r\n\r\nFixed an issue in `config\u002Fgenerate_mcp_yaml.py` where the script failed to correctly parse MCP tool names containing hyphens (like those from the Context7 server). This prevented these tools from being included in the generated MCP server list used by the AI. The fix updates the parsing logic to correctly handle hyphens in tool names, ensuring all available MCP tools are properly listed. Thanks to [Josses92](https:\u002F\u002Fgithub.com\u002FJosses92) for pointing it out!\r\n","2025-05-04T19:31:52",{"id":163,"version":164,"summary_zh":165,"released_at":166},106621,"v0.4.5","# RooFlow v0.4.5\r\n\r\n**MCP Server Tool Usage Added To RooFlow Prompts!**\r\n\r\n\u003Cbr>\r\n\r\n# RooFlow v0.4.4\r\n\r\n**Revised and streamlined the mode instructions and added Orchestrator mode definition.**\r\n\r\n\u003Cbr>\r\n\r\n# RooFlow v0.4.3 - \r\n**Addressing Line Count Issues During File Modification:**\r\n\r\n1.  **Corrected Tool Usage Formats and Examples:** I noticed that the `usage_format` and `example.usage` for several tools (`read_file`, `fetch_instructions`, `search_files`, `list_files`, `list_code_definition_names`, `write_to_file`, `attempt_completion`, `switch_mode`, `new_task`, `execute_command`) were using a YAML-like format (`tool_name: { param: value }`) instead of the required XML format (`\u003Ctool_name>\u003Cparam>value\u003C\u002Fparam>\u003C\u002Ftool_name>`) as defined in the `tool_use_protocol`. **This is a critical inconsistency that needs fixing.** I have corrected all of these to use the XML format.\r\n2.  **Corrected `apply_diff` Diff Format:** The `apply_diff` tool definition's `diff` format block was missing `:end_line:`, but the parameter description and examples included it. I've corrected the format block definition to include `:end_line:` for consistency.\r\n3.  **New Rule R13 (File Edit Preparation):** Added a specific rule under `rules` that explicitly states that *before* using `apply_diff` or `insert_content` on an *existing* file, the AI *must* obtain the file's content *with line numbers* using `read_file`, unless that information is already available (e.g., from user-provided content). This makes `read_file` a mandatory preceding step for these edits.\r\n4.  **New Rule R14 (File Edit Error Recovery):** Added a rule under `rules` that provides a specific procedure for when *any* file modification tool fails. The first step in recovery is *always* to `read_file` on the target file to get its current state before attempting to fix the issue and retry.\r\n5.  **Updated Objective Step 3 (Analysis and Planning):** Modified the planning steps (specifically added step 'e') to include a check for the required file content with line numbers *if* the chosen tool is `apply_diff` or `insert_content` on an existing file. This integrates the R13 rule into the core planning logic.\r\n6.  **Refined `read_file` Description:** Slightly enhanced the `read_file` description to explicitly mention its output includes line numbers and its purpose for planning modifications.\r\n7.  **Refined `write_to_file` `line_count` Description:** Clarified that `line_count` is based on the *provided content*.\r\n8.  **Refined `insert_content` Example:** Fixed indentation in the multi-line content example and added a `pass` statement to make it valid Python syntax.\r\n\r\nThese changes make the requirement for obtaining file content and line numbers via `read_file` much more explicit as a necessary step before attempting certain file modifications, and provide a clear recovery path if those modifications fail. The correction of the XML usage format across multiple tools is also essential for the prompt's consistency and the AI's ability to call tools correctly.\r\n\r\n# RooFlow v0.4.2\r\n**Improvements for AI Clarity and Usability:**\r\n\r\n1.  **Structured YAML:** Standard format for easy parsing.\r\n2.  **Clearer Description:** Combined descriptive points, explicitly stated the purpose (surgical replace), the benefit (maintains formatting), and the rule about consolidating multiple changes into one call. Added base path context.\r\n3.  **Precise Parameter Definition (`diff`):**\r\n    *   Used a multi-line string (`|`) to clearly lay out the exact required format of a diff block.\r\n    *   **Explicitly noted that `:end_line:` is NOT used** in this version of the format, resolving ambiguity with previous versions.\r\n    *   **Prominently featured the \"IMPORTANT ESCAPING RULE\"** within the parameter description itself, detailing *which* markers need escaping (`\u003C\u003C\u003C\u003C\u003C\u003C\u003C SEARCH`, `=======`, `>>>>>>> REPLACE`) and *how* (`\\`) when they appear *as content* within the SEARCH or REPLACE sections. This directly addresses the common error scenario.\r\n    *   Reiterated the rule about the single structural `=======`.\r\n4.  **Structured Examples:** Presented examples clearly with descriptions and the exact `usage` format.\r\n5.  **Escaping Example:** Added a dedicated example demonstrating the critical escaping rule, which is vital for the AI to correctly handle files containing merge conflict markers or similar text.\r\n6.  **Consistency:** Ensured the base path (`\u002Fvar\u002Fwww\u002Froo-flow`) is consistent with the provided markdown.\r\n\r\nThis revised YAML provides a much more robust and less error-prone instruction set for the AI when using the `apply_diff` tool.\r\n\u003Cbr>\r\n\r\n# RooFlow v0.4.1\r\n\r\n**Updated the RooFlow system prompts with the latest Roo Code tool definitions.**\r\n\u003Cbr>\r\n\r\n# RooFlow v0.4.0\r\n\r\n* **RooFlow is now installed alongside the standard Roo Code modes, as Flow-Architect, Flow-Code, Flow-Debug , and Flow-Ask.**\r\n* **Now you can easily switch to a standard Roo Code mode if something is not working correctly","2025-03-03T16:43:14"]