[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-73271":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":23,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":41,"readmeContent":42,"aiSummary":43,"trendingCount":16,"starSnapshotCount":16,"syncStatus":44,"lastSyncTime":45,"discoverSource":46},73271,"wanwu","UnicomAI\u002Fwanwu","UnicomAI","China Unicom's Yuanjing Wanwu Agent Platform is an enterprise-grade, multi-tenant AI agent development platform. It helps users build applications such as intelligent agents, workflows, and rag, and also supports model management. The platform features a developer-friendly license, and we welcome all developers to build upon the platform.","",null,"Go",2526,109,31,15,0,3,6,21,9,28.12,"Apache License 2.0",false,"main",[26,27,28,29,30,31,32,33,34,35,36,37,38,39,5,40],"agent","agentic-ai","agentic-framework","ai","ai-agent","ai-agent-development-framework","ai-agents-framework","development","genai","golang","llm","mcp","open-ai","rag","workflow","2026-06-12 02:03:11","\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F4788ed8f-eefc-4c19-aa77-7ec776743f3d\" style=\"width:45%; height:auto;\" \u002F>\n\u003Cp>\n  \u003Ca href=\"#🚩 Core Function Modules\">Core Function Modules\u003C\u002Fa> •\n  \u003Ca href=\"#x1F3AF; Typical Application Scenarios\">Typical Application Scenarios\u003C\u002Fa> •\n  \u003Ca href=\"#🚀 Quick Start\">Quick Start\u003C\u002Fa> •\n  \u003Ca href=\"#x1F4D1; Using Wanwu\">Using Wanwu\u003C\u002Fa> •\n  \u003Ca href=\"#128172; Q & A\">Q & A\u003C\u002Fa> •\n  \u003Ca href=\"#x1F4E9; Contact Us\">Contact Us\u003C\u002Fa> \n\u003C\u002Fp>\n\u003Cp>\n  \u003Cimg alt=\"License\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-apache2.0-blue.svg\">\n  \u003Cimg alt=\"Go Version\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fgo-%3E%3D%201.24.0-blue\">\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Freleases\">\n    \u003Cimg alt=\"Release Notes\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002FUnicomAI\u002Fwanwu?label=Release&logo=github&color=green\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n    English |\n    \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002FREADME_CN.md\">简体中文\u003C\u002Fa> |\n    \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002FREADME_繁體.md\">繁體中文\u003C\u002Fa>\n\u003C\u002Fp>\n\u003C\u002Fdiv>\n\n\nThe **Yuanjing Wanwu AI Agent Platform** is an **enterprise-oriented**, **one-stop**, and **commercial-license-friendly** **agent development platform**, dedicated to providing enterprises with secure, efficient, and compliant one-stop AI solutions. With the core philosophy of \"open technology and collaborative ecosystem building,\" we integrate cutting-edge technologies such as large language models and business process automation to build an AI engineering platform with a complete functional system covering full-lifecycle model management, MCP, web search, **General agent & Skills orchestration**, enterprise knowledge base construction, and complex workflow orchestration. The platform has now fully upgraded to a **\"General Agent + Vertical Scenario Skills\" dual-engine development platform**. While ensuring enterprise data security and privacy protection, it significantly lowers the application threshold of AI technology, helping enterprises accelerate their digital transformation processes to achieve cost reduction, efficiency enhancement, and business innovation.\n\n------\n\n\u003Cdiv>\n  \u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fwww.bilibili.com\u002Fvideo\u002FBV1HxpazNEAM\">\u003Cimg width=\"400\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F54efe5d3-c28d-48fb-9a6e-d6ac536a1f95\" \u002F>\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.bilibili.com\u002Fvideo\u002FBV1HxpazNEAM\">\u003Cimg width=\"394\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fd19831e6-10a3-4ee0-8caf-6c0ebe2af4a5\" \u002F>\u003C\u002Fa>\n  \u003C\u002Fp>\n\u003C\u002Fdiv>\n\n------\n\n### 📢 Open Ecosystem\n\n- [External Knowledge Base Compatibility](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F2.%E7%9F%A5%E8%AF%86%E5%BA%93\u002F%E8%BF%9E%E6%8E%A5%E5%A4%96%E6%8E%A5%E7%9F%A5%E8%AF%86%E5%BA%93.md): Supports API-based import of knowledge bases created in Dify, with retrieval and recall in agents, Q&A, and workflows.\n- [MCP Hub](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F2.%E8%B5%84%E6%BA%90%E5%BA%93%2FMCP%E6%9C%8D%E5%8A%A1.md): Supports importing and using MCP from different service providers.\n- [Skills](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F2.%E8%B5%84%E6%BA%90%E5%BA%93%2FSkills.md): Supports creating and downloading Skills, with seamless integration to OpenClaw.\n- [OpenClaw Sandbox](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F8.%E9%80%9A%E7%94%A8%E6%99%BA%E8%83%BD%E4%BD%93%2F%E6%9C%BA%E5%99%A8%E4%BA%BA%E5%8A%A9%E6%89%8B-OPENCLAW%2F%E5%A6%82%E4%BD%95%E5%9C%A8%E4%B8%87%E6%82%9F%E4%B8%AD%E6%8E%A5%E5%85%A5OpenClaw%E6%9C%BA%E5%99%A8%E4%BA%BA.md): We provide the option to deploy each “OpenClaw Robot” in a standalone Docker container. You can directly access your locally deployed OpenClaw robot within Yuanjing Wanwu.\n\n------\n\n### 🔥 Adopt a permissive and friendly Apache 2.0 License, supporting developers to freely expand and develop secondary\n\n✔ **Enterprise-level engineering**: Provides a complete toolchain from model management to application landing, solving the \"last mile\" problem of LLM technology landing\n\n✔ **Open-source ecological**: Adopt a permissive and friendly **Apache 2.0 License**, supporting developers to freely expand and develop\n\n✔ **Full-stack technology support**: Equipped with a professional team to provide **architecture consulting, performance optimization** and full-cycle empowerment for ecological partners\n\n✔ **Multi-tenant architecture**: Provides a multi-tenant account system to meet the core needs of users in cost control, data security isolation, business elasticity expansion, industry customization, rapid online and ecological collaboration\n\n✔ **XinChuang adaptation**: The product has been awarded the **“Xinchuang AI Hardware and Software System Inspection Certificate“**，featuring hardware support for Huawei Kunpeng CPUs and software compatibility with domestic operating systems (e.g., openEuler, CULinux, Kylin) and databases (e.g., TiDB, OceanBase).\n\n------\n\n### 🚩 Core Function Modules\n\n#### **1. Model Management (Model Hub)**\n▸ Supports the unified access and lifecycle management of **hundreds of proprietary\u002Fopen-source large models** (including GPT, Claude, Llama, etc.)\n\n▸ Deeply adapts to **OpenAI API standards** and **Unicom Yuanjing** ecological models, realizing seamless switching of heterogeneous models\n\n▸ Provides **multi-inference backend support** (vLLM, TGI, etc.) and **self-hosted solutions** to meet the computing power needs of enterprises of different scales\n\n#### **2. MCP**\n▸ **Standardized interfaces**: Enable AI models to seamlessly connect to various external tools (such as GitHub, Slack, databases, etc.) without the need to develop adapters for each data source separately\n\n▸ **Built-in rich and selected recommendations**: Integrates 100+ industry MCP interfaces, making it easy for users to call up quickly and easily\n\n#### **3. Web Search**\n▸ **Real-time information acquisition**: Possesses powerful web search capabilities, capable of obtaining the latest information from the Internet in real-time. In question and answer scenarios, when a user's question requires the latest news, data, and other information, the platform can quickly search and return accurate results, enhancing the timeliness and accuracy of the answers\n\n▸ **Multi-source data integration**: Integrates various Internet data sources, including news websites, academic databases, industry reports, etc. Through the integration and analysis of multi-source data, it provides users with more comprehensive and in-depth information. For example, in market research scenarios, relevant data can be obtained from multiple data sources at the same time for comprehensive analysis and evaluation\n\n▸ **Intelligent search strategy**: Adopt intelligent search algorithms, automatically optimize search strategies based on user questions to improve search efficiency and accuracy. Support keyword search, semantic search and other search methods to meet the needs of different users. At the same time, intelligently sort and filter search results, prioritize the display of the most relevant and valuable information\n\n#### **4. Visual Workflow (Workflow Studio)**\n▸ Quickly build complex AI business processes through **low-code drag-and-drop canvas**\n\n▸ Built-in **conditional branching, API, large model, knowledge base, code, MCP** and other nodes, support end-to-end process debugging and performance analysis\n\n#### 5. \u003Ca href=\"#🚀High-precision RAG\">High-precision RAG\u003C\u002Fa>\n▸ Provides the whole process knowledge management capabilities of **knowledge base creation → document parsing → vectorization → retrieval → fine sorting**, supports **multiple formats** such as pdf\u002Fdocx\u002Ftxt\u002Fxlsx\u002Fcsv\u002Fpptx documents, and also supports the capture and access of web resources\n\n▸ Integrates **multi-modal retrieval**, **cascading segmentation** and **adaptive segmentation**, significantly improves the accuracy of Q&A\n\n#### **6. General Agent & Skills Orchestration Framework** \n\n▸ **Dual-Engine Mode**: Breaks the limitation of traditional agents \"having a brain but no hands,\" upgrading to a \"General Agent + Vertical Scenario Skills\" dual-engine platform to create an enterprise-level super agent that is both \"knowledgeable\" and \"professional\" \n\n▸ **Almighty Brain**: The general agent, as the core engine, has now demonstrated professional analyst-level multi-step reasoning and information integration capabilities in complex scenarios like deep research and data analysis \n\n▸ **Minimalist Skill Building**: Supports **\"one-sentence Skill creation\"**. No code is required; just describe your needs in natural language to automatically generate vertical scenario skills, turning business experience into a dedicated \"toolbox\" \n\n▸ **Zero-Code Orchestration Closed Loop**: The **industry's first to support zero-code Skill calling during agent development**. Directly associate Skills in the visual interface to achieve a perfect closed loop from \"intent recognition\" to \"skill execution\" \n\n▸ **On-Demand Toolbox**: Flexibly configure and call built-in tools, Skills, MCP, workflows, and other agents, making AI not only \"think\" but also \"act\" ▸ **Read Hundreds of Pages in Seconds**: Supports uploading various files; the general agent can quickly parse them and conduct precise deep Q&A and interaction based on the files \n\n▸ **Unified Workspace**: Provides a unified destination for outcomes, neatly displaying all interactively generated files with support for online preview and one-click download \n\n▸ **Basic Development Paradigm**: Still supports traditional Agent construction based on **function calling**, supporting private knowledge base association and multi-round online debugging\n\n#### 7.Wanwu Ontology Agent\n\n▸ Automatically constructs business knowledge networks from enterprise data and documents, empowering AI with deep reasoning and closed-loop action capabilities to truly understand business and make decisions.\n\n#### **8. Backend as a Service (BaaS)**\n▸ Provides **RESTful API**, supports deep integration with existing enterprise systems (OA\u002FCRM\u002FERP, etc.)\n\n▸ Provides **fine-grained permission control** to ensure stable operation in production environments\n\n------\n\n### 📢 Function Comparison\n|                    Function                    | Wanwu |             Dify.AI             |          Fastgpt           |             Ragflow             |    Coze open source version     |\n| :--------------------------------------------: | :---: | :-----------------------------: | :------------------------: | :-----------------------------: | :-----------------------------: |\n|                  Model import                  |   ✅   |                ✅                |     ❌(Built-in models)     |                ✅                |       ❌(Built-in models)        |\n|               Direct OCR import                |   ✅   |                ❌                |             ❌              |                ❌                |                ❌                |\n|                   RAG engine                   |   ✅   |                ✅                |             ✅              |                ✅                |                ✅                |\n|                    GraphRAG                    |   ✅   |                ❌                |             ❌              |                ✅                |                ❌                |\n|                 Ontology Agent                 |   ✅   |                ❌                |             ❌              |                ❌                |                ❌                |\n|    Multi-Agent Orchestration & Development     |   ✅   |                ❌                |             ✅              |                ✅                |                ❌                |\n| General Agent & Skills Orchestration Framework |   ✅   |                ❌                |             ❌              |                ❌                |                ❌                |\n|                     Agent                      |   ✅   |                ✅                |             ✅              |                ✅                |                ✅                |\n|                    Workflow                    |   ✅   |                ✅                |             ✅              |                ✅                |                ✅                |\n|                      MCP                       |   ✅   |                ✅                |             ✅              | ✅(Need to install tools to use) |                ❌                |\n|               Search enhancement               |   ✅   | ✅(Need to install tools to use) |             ✅              | ✅(Need to install tools to use) |                ✅                |\n|                Local deployment                |   ✅   |                ✅                |             ✅              |                ✅                |                ✅                |\n|                  Multi-tenant                  |   ✅   |   ❌(Commercially restricted)    | ❌(Commercially restricted) |                ✅                | ✅(Users are not interconnected) |\n|                license friendly                |   ✅   |   ❌(Commercially restricted)    | ❌(Commercially restricted) |      Not fully open source      |                ✅                |\n> As of May 15, 2026.\n\n------\n\n### 🎯 Typical Application Scenarios\n\n- **Intelligent Customer Service**: Realize high-accuracy business consultation and ticket processing based on RAG + Agent\n- **Knowledge Management**: Build an exclusive enterprise knowledge base, support semantic search and intelligent summary generation\n- **Process Automation**: Realize AI-assisted decision-making for business processes such as contract review and reimbursement approval through the workflow engine\n\nThe platform has been successfully applied in multiple industries such as **finance, industry, and government**, helping enterprises transform the theoretical value of LLM technology into actual business benefits. We sincerely invite developers to join the open source community and jointly promote the democratization of AI technology.\n\n------\n\n### 🚀 Quick Start\n\n- The workflow module of the Wanwu AI Agent Platform uses the following project, you can go to its warehouse to view the details.\n  - v0.1.8 and earlier: wanwu-agentscope project\n  - v0.2.0 and later: [wanwu-workflow](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu-workflow\u002Ftree\u002Fdev\u002Fwanwu-backend) project\n\n- **Recommended Configuration:**\n  - CPU: 8-core or 16-core; RAM: 32GB; Storage: 200GB or more; GPU: Not required.\n\n- **Model Requirements:**\n  - When using WanwuBot (General Agent) or creating Skills with a single command, the selected model must have a context length >= 32000 when importing.\n  \n- **Docker Installation (Recommended)**\n\n1. Before the first run\n\n    1.1 Copy the environment variable file\n    ```bash\n    cp .env.example .env\n    ```\n\n    1.2 Modify the `WANWU_ARCH` and `WANWU_EXTERNAL_IP` variables in the .env file according to the system\n    ```\n    # amd64 \u002F arm64\n    WANWU_ARCH=amd64\n    \n    # external ip port (Note: if the browser accesses Wanwu deployed on a non-localhost server, you need to change localhost to the external IP, for example, 192.168.xx.xx)\n    WANWU_EXTERNAL_IP=localhost\n    ```\n\n    1.3 Configure the `WANWU_BFF_JWT_SIGNING_KEY` variable in the .env file, a custom complex random string used for generating JWT tokens\n    ```\n    # bff\n    WANWU_BFF_JWT_SIGNING_KEY=\n    ```\n\n    1.4 Create a Docker running network\n    ```\n    docker network create wanwu-net\n    ```\n\n2. Start the service (the image will be automatically pulled from Docker Hub during the first run)\n\n    ```bash\n    # For amd64 system:\n    docker compose --env-file .env --env-file .env.image.amd64 up -d\n    # For arm64 system:\n    docker compose --env-file .env --env-file .env.image.arm64 up -d\n    ```\n\n3. Log in to the system: http:\u002F\u002Flocalhost:8081\n\n    ```\n    Default user: admin\n    Default password: Wanwu123456\n    ```\n\n4. Stop the service\n\n    ```bash\n    # For amd64 system:\n    docker compose --env-file .env --env-file .env.image.amd64 down\n    # For arm64 system:\n    docker compose --env-file .env --env-file .env.image.arm64 down\n    ```\n\n5. Having trouble pulling middleware or other Docker images? We've prepared a backup of the images on Netdisk. Please follow the instructions in its README file: [Wanwu Docker Image Backup](https:\u002F\u002Fpan.baidu.com\u002Fe\u002F1cupIcEP2RBwi_hOr4xQnFQ?pwd=ae86)\n\n- **Source Code Start (Development)**\n\n1. Based on the above Docker installation steps, start the system service completely\n\n2. Take the backend bff-service service as an example\n\n    2.1 Stop bff-service\n    ```\n    make -f Makefile.develop stop-bff\n    ```\n\n    2.2 Compile the bff-service executable file\n    ```\n    # For amd64 system:\n    make build-bff-amd64\n    # For arm64 system:\n    make build-bff-arm64\n    ```\n\n    2.3 Start bff-service\n    ```\n    make -f Makefile.develop run-bff\n    ```\n\n------\n\n### ⬆️ Version Upgrade\n\n1. Based on the above Docker installation steps, completely stop the system service\n\n2. Update to the latest version of the code\n\n    2.1 In the wanwu repository directory, update the code\n    ```bash\n    # Switch to the main branch\n    git checkout main\n    # Pull the latest code\n    git pull\n    ```\n\n    2.2 Recopy the environment variable file (if there are changes to the environment variables, please modify them again)\n    ```bash\n    # Backup the current .env file\n    cp .env .env.old\n    # Copy the .env file\n    cp .env.example .env\n    ```\n\n3. Based on the above Docker installation steps, completely start the system service\n\n------\n\n### 🧬 Start Ontology Agent Platform\n\n1. Based on the above Docker installation steps, completely start the system service\n\n2. Before the first run\n\n    2.1 Generate RSA key pair\n    ```bash\n    .\u002Fconfigs\u002Fmicroservice\u002Fontology\u002Fvega-server\u002Fgenerate-keys.sh configs\u002Fmicroservice\u002Fontology\u002Fvega-server\n    ```\n\n    2.2 Generate frontend public key configuration (cross-platform, requires Node environment)\n    ```bash\n    node configs\u002Fmicroservice\u002Fontology\u002Fvega-server\u002Fgenerate-public-key-js.js\n    ```\n\n3. Copy environment variable file (before first run or after system upgrade)\n\n    ```bash\n    # Backup current .env.ontology file (if exists)\n    cp .env.ontology .env.ontology.old\n    # Copy .env.ontology file\n    cp .env.ontology.example .env.ontology\n    ```\n\n4. Start the service\n\n    4.1 Confirm ontology feature is enabled in .env file\n    ```\n    WANWU_BFF_ONTOLOGY_ENABLE=1\n    ```\n\n    4.2 Start ontology agent service\n    ```bash\n    # For amd64 system:\n    docker compose --env-file .env --env-file .env.ontology --env-file .env.image.amd64 -f docker-compose.ontology.yaml up -d\n    # For arm64 system:\n    docker compose --env-file .env --env-file .env.ontology --env-file .env.image.arm64 -f docker-compose.ontology.yaml up -d\n    ```\n\n5. Stop the service\n    ```bash\n    # For amd64 system:\n    docker compose --env-file .env --env-file .env.ontology --env-file .env.image.amd64 -f docker-compose.ontology.yaml down\n    # For arm64 system:\n    docker compose --env-file .env --env-file .env.ontology --env-file .env.image.arm64 -f docker-compose.ontology.yaml down\n    ```\n\n------\n\n### ➡️ Xinchuang Adaptation (TiDB & OceanBase)\n\n1. Based on the above Docker installation steps, complete step before the first run\n\n2. Modify the `WANWU_DB_NAME` variable in the .env file according to the database\n\n3. Start the database (taking amd64 as an example)\n   ```bash\n   # tidb\n   docker compose --env-file .env --env-file .env.image.amd64 -f docker-compose.tidb.yaml up -d\n   # oceanbase\n   docker compose --env-file .env --env-file .env.image.amd64 -f docker-compose.oceanbase.yaml up -d\n   ```\n\n4. Based on the above Docker installation steps, completely start the system service\n\n✔ The product has been awarded the “Xinchuang AI Hardware and Software System Inspection Certificate,” featuring hardware support for Huawei Kunpeng CPUs and software compatibility with domestic operating systems (e.g., openEuler, CULinux, Kylin) and databases (e.g., TiDB, OceanBase).\n\n------\n\n### 📑 Using Wanwu\nTo help you quickly get started with this project, we strongly recommend that you first check out the [ Documentation Operation Manual](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Ftree\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual). We provide users with interactive and structured operation guides, where you can directly view operation instructions, interface documents, etc., greatly reducing the threshold for learning and use. The detailed function list is as follows:\n\n| Feature                                                      | Detailed Description                                         |\n| :----------------------------------------------------------- | :----------------------------------------------------------- |\n| [General Agent](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Ftree\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F8.%e9%80%9a%e7%94%a8%e6%99%ba%e8%83%bd%e4%bd%93) | The platform deeply integrates advanced capabilities such as deep research and data analysis, achieving a comprehensive leap from simple Q&A to complex business processing, creating your all-around AI digital assistant. |\n| Ontology Agent                                               | Automatically constructs business knowledge networks from enterprise data and documents, empowering AI with deep reasoning and closed-loop action capabilities to truly understand business and make decisions. |\n| [Model Management](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F1.%E6%A8%A1%E5%9E%8B%E7%AE%A1%E7%90%86.md) | Supports users to import LLM, Embedding, and Rerank models from various model providers, including Unicom Yuanjing, OpenAI-API-compatible, Ollama, Tongyi Qianwen, and Volcano Engine. [Model Import Methods - Detailed Version](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F%E6%A8%A1%E5%9E%8B%E5%AF%BC%E5%85%A5%E6%96%B9%E5%BC%8F-%E8%AF%A6%E7%BB%86%E7%89%88.md) |\n| [Knowledge Base](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Ftree\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F2.%E7%9F%A5%E8%AF%86%E5%BA%93) | In terms of document parsing capabilities: supports uploading of 12 file types and URL parsing; Supports private deployment and integration for document parsing via two methods: OCR and [a proprietary MinerU model (for scenarios like titles, tables, and formulas)](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002FDocParserServer\u002Ftree\u002Fmain) ; document segmentation settings support both general segmentation and parent-child segmentation. In terms of optimization capabilities: supports metadata management 、Graph RAG and metadata filtering queries, supports adding, deleting, and modifying segmented content, supports setting keyword tags for segments to improve recall performance, supports segment enable\u002Fdisable operations, and supports hit testing. In terms of retrieval capabilities: supports multiple retrieval modes including vector search, full-text search, and hybrid search. In terms of Q&A capabilities: supports automatic citation of sources and generating answers with both text and images.\u003Cbr |\n| [Resource Library](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F3.%E5%B7%A5%E5%85%B7%E5%B9%BF%E5%9C%BA.md) | Supports importing your own MCP services or custom tools or skills for use in workflows and agents. |\n| [Safety Guardrails](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F4.%E5%AE%89%E5%85%A8%E6%8A%A4%E6%A0%8F.md) | Users can create sensitive word lists to control the safety of the model's output. |\n| [Text Q&A](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F5.%E6%96%87%E6%9C%AC%E9%97%AE%E7%AD%94.md) | A dedicated knowledge advisor based on a private knowledge base. It supports features like knowledge base management, Q&A, knowledge summarization, personalized parameter configuration, safety guardrails, and retrieval configuration to improve the efficiency of knowledge management and learning. Supports publishing text Q&A applications publicly or privately, and can be published as an API. |\n| [Workflow](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Ftree\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F6.%E5%B7%A5%E4%BD%9C%E6%B5%81) | Extends the capabilities of agents. Composed of nodes, it provides a visual workflow editor. Users can orchestrate multiple different workflow nodes to implement complex and stable business processes. Supports publishing workflow applications publicly or privately, can be published as an API, and supports import\u002Fexport. |\n| [Agent](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F7.%E6%99%BA%E8%83%BD%E4%BD%93.md) | Create agents based on user scenarios and business requirements. Supports model selection, prompt setting, web search, knowledge base selection, MCP, workflows, and custom tools. Supports publishing agent applications publicly or privately, and can be published as an API and a Web URL. |\n| [App Marketplace](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F8.%E5%BA%94%E7%94%A8%E5%B9%BF%E5%9C%BA.md) | Allows users to experience published applications, including Text Q&A, Workflows, and Agents. |\n| [MCP Hub](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F9.MCP%E5%B9%BF%E5%9C%BA.md) | Features 100+ pre-selected industry-specific MCP servers, ready for immediate use. |\n| [Template Plaza](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F10.%E6%A8%A1%E6%9D%BF%E5%B9%BF%E5%9C%BA.md) | Built-in with 50+ optimized industry prompts, available for immediate use. |\n| [Settings](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002Fmain\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F9.%E8%AE%BE%E7%BD%AE.md) | The platform supports multi-tenancy, allowing users to manage organizations, roles, users, and perform basic platform configuration. |\n| [UniAI-GraphRAG](https:\u002F\u002Fgithub.com\u002FUnicomAI\u002Fwanwu\u002Fblob\u002F66539378255f9a1da80b02a83e75c7a5155f7f87\u002Fconfigs\u002Fmicroservice\u002Fbff-service\u002Fstatic\u002Fmanual\u002F2.%E7%9F%A5%E8%AF%86%E5%BA%93\u002F%E5%88%9B%E5%BB%BA%E7%9F%A5%E8%AF%86%E5%BA%93%E3%80%81%E9%97%AE%E7%AD%94%E5%BA%93\u002F%E5%88%9B%E5%BB%BA%E7%9F%A5%E8%AF%86%E5%BA%93\u002F%E7%9F%A5%E8%AF%86%E5%9B%BE%E8%B0%B1%E4%BD%BF%E7%94%A8%E8%AF%B4%E6%98%8E.md) | UniAI-GraphRAG integrates techniques such as domain knowledge ontology modeling, knowledge graph and community report construction, and Graph Retrieval-Augmented Generation to effectively enhance the completeness, logical coherence, and credibility of knowledge question answering. It significantly improves performance in complex QA scenarios like cross-document summarization and multi-hop relational reasoning. |\n\n### 🚀High-precision RAG\n\n**Wanwu RAG has completed its retrieval performance evaluation on the authoritative, publicly available industry benchmark, the MultiHop-RAG dataset**\n\n\u003Cp align=\"center\">\n  \u003Cimg width=\"584\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F8a267ba2-13e4-48fe-8ea8-4f24fb10dfc6\" \u002F>\n\u003C\u002Fp>\n\nThe F1 score serving as the comprehensive evaluation metric (the harmonic mean of precision and recall), are as follows: \n\n1）Wanwu RAG outperforms Dify by 14% \n\n2）Wanwu GraphRAG outperforms Dify by 17.2% \n\n3）Wanwu GraphRAG outperforms open-source LightRAG by 3.5%\n\n------\n\n### 📰 TO DO LIST\n\n- [x] General Agent\n- [x] Skills\n- [ ] Support importing databases into knowledge base\n- [ ] A2A Protocol\n- [ ] Agent and Model Evaluation\n- [ ] Trace Tracking\n\n------\n\n### 💬 Q & A\n\n- **[Q] Error when starting Elastic (elastic-wanwu) on Linux system: Memory limited without swap.**\n  **[A]** Stop the service, run `sudo sysctl -w vm.max_map_count=262144`, and then restart the service.\n  \n- **[Q] After the system services start normally, the mysql-wanwu-setup and elastic-wanwu-setup containers exit with status code Exited (0).**\n  **[A]** This is normal. These two containers are used to complete some initialization tasks and will automatically exit after execution.\n  \n- **[Q] Regarding model import**\n  **[A]** Taking the import of Unicom Yuanjing LLM as an example (the process is similar for importing OpenAI-API-compatible models, Embedding, or Rerank types):\n  ```\n  1. The Open API interface for Unicom Yuanjing MaaS Cloud LLM is, for example: https:\u002F\u002Fmaas.ai-yuanjing.com\u002Fopenapi\u002Fcompatible-mode\u002Fv1\u002Fchat\u002Fcompletions\n  2. The API Key applied for by the user on Unicom Yuanjing MaaS Cloud looks like: sk-abc********************xyz\n  3. Confirm that the API and Key can correctly request the LLM. Taking a request to yuanjing-70b-chat as an example:\n      curl --location 'https:\u002F\u002Fmaas.ai-yuanjing.com\u002Fopenapi\u002Fcompatible-mode\u002Fv1\u002Fchat\u002Fcompletions' \\\n      --header 'Content-Type: application\u002Fjson' \\\n      --header 'Accept: application\u002Fjson' \\\n      --header 'Authorization: Bearer sk-abc********************xyz' \\\n      --data '{\n              \"model\": \"yuanjing-70b-chat\",\n              \"messages\": [{\n                      \"role\": \"user\",\n                      \"content\": \"你好\"\n              }]\n      }'\n  4. Import the model:\n  4.1 [Model Name] must be the model that can be correctly requested in the curl command above; for example, yuanjing-70b-chat.\n  4.2 [API Key] must be the key that can be correctly requested in the curl command above; for example, sk-abc********************xyz (note: do not include the 'Bearer' prefix).\n  4.3 [Inference URL] must be the URL that can be correctly requested in the curl command above; for example, https:\u002F\u002Fmaas.ai-yuanjing.com\u002Fopenapi\u002Fcompatible-mode\u002Fv1 (note: do not include the \u002Fchat\u002Fcompletions suffix).\n  5. Importing an Embedding model is the same as importing an LLM as described above. Note that the inference URL should not include the \u002Fembeddings suffix.\n  6. Importing a Rerank model is the same as importing an LLM as described above. Note that the inference URL should not include the \u002Frerank suffix.\n  ```\n\n------\n\n### 🔗 Acknowledgments\n\n- [Coze](https:\u002F\u002Fgithub.com\u002Fcoze-dev)\n- [LangChain](https:\u002F\u002Fgithub.com\u002Flangchain-ai\u002Flangchain)\n- [AIO Sandbox](https:\u002F\u002Fgithub.com\u002Fagent-infra\u002Fsandbox)\n- [OpenCode](https:\u002F\u002Fgithub.com\u002Fanomalyco\u002Fopencode)\n- [KWeaver Core](https:\u002F\u002Fgithub.com\u002Fkweaver-ai\u002Fkweaver-core)\n\n------\n\n### ⚖️ License\nThe Yuanjing Wanwu AI Agent Platform is released under the Apache License 2.0.\n\n------\n\n### 📩 Contact Us\n| QQ Group1(Full):490071123                                    | QQ Group2(Full):1026898615                                         | QQ Group3:1019579243                                         |\n| ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |\n| \u003Cimg width=\"183\" height=\"258\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F163d6580-af84-4fe4-9b51-7effb4153dd8\" \u002F> | \u003Cimg width=\"183\" height=\"258\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F03d10f7c-7460-485e-9f17-b3135d460dd0\" \u002F> | \u003Cimg width=\"183\" height=\"258\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F6cf67753-899c-418d-971b-f43fc9b5bada\" \u002F> |","中国联通的元境万物AI代理平台是一个面向企业的一站式多租户AI代理开发平台，旨在帮助企业构建智能代理、工作流及RAG应用，并支持模型管理。该平台采用Go语言编写，具备全生命周期模型管理、MCP、网络搜索等功能模块，支持通用代理与技能编排、企业知识库建设及复杂工作流编排。其“通用代理+垂直场景技能”双引擎开发模式在保障企业数据安全和隐私保护的同时，显著降低了AI技术的应用门槛，适用于需要加速数字化转型以实现降本增效和业务创新的企业场景。",2,"2026-06-11 03:44:46","high_star"]