A modular, extensible Retrieval-Augmented Generation (RAG) platform built with LangChain 1.x, Chroma, and DashScope LLMs.
Overview RAG Platform is a production-oriented AI agent framework that combines retrieval (RAG) and reasoning (Agent) into a unified system. It is designed for rapid development of intelligent applications such as: AI customer service systems Recommendation engines Knowledge-based QA systems Automated report generation Prompt optimization for generative AI The project emphasizes modularity, scalability, and reusability, making it suitable for both academic projects and real-world applications.
🖥️ Demo Preview
Features 🤖 Multi-Agent System Create and manage multiple agents Agent-level configuration isolation Dynamic agent switching Independent conversation context per agent 📚 Knowledge Base (RAG) Supports TXT / PDF ingestion Automatic document chunking Embedding via DashScope Vector storage using Chroma Knowledge isolation per agent 💬 Intelligent Chat ReAct-based agent architecture Tool calling support Context-aware conversations Session management with reset capability 🛠 Tooling System Built-in tools include: rag_summarize – retrieval + summarization get_user_id get_current_month fetch_external_data fill_context_for_report Easily extensible with custom tools. 📊 Observability Tool execution logging Prompt debugging support Structured runtime logs
Architecture User Input ↓ Streamlit UI ↓ Agent Service ↓ LangChain Agent (ReAct) ↓ Tools / RAG Retrieval ↓ Vector Store (Chroma) ↓ Embedding Model (DashScope) ↓ LLM (ChatTongyi) ↓ Response
Project Structure rag_platform/ ├── app.py ├── services/ ├── rag/ ├── tools/ ├── ui/ ├── utils/ ├── config/ ├── data/ ├── chroma_db/ └── requirements.txt
Installation
-
Clone the repository git clone https://github.com/your-username/rag-platform.git cd rag-platform
-
Create virtual environment python -m venv .venv
-
Activate environment .venv\Scripts\activate # Windows
-
Install dependencies pip install -r requirements.txt
-
Install DashScope SDK pip install dashscope
-
Set API Key set DASHSCOPE_API_KEY=your_api_key
Usage streamlit run app.py Open browser: http://localhost:8501
Core Components AgentService
Manages agent lifecycle, configuration, and switching.
RagService
Handles document retrieval, context construction, and LLM interaction.
VectorStoreService
Responsible for embedding, storage, and retrieval.
Tool System
Encapsulates external capabilities and integrates with ReAct workflow.
🔧 Design Principles Decoupling: Agent logic separated from RAG pipeline Config-driven: YAML-based configuration Extensibility: Plug-and-play tools and models Modularity: Clean layered architecture 🚀 Roadmap
💡 Use Cases Enterprise knowledge assistant AI recommendation systems Educational AI tools Data analysis agents Prompt engineering assistants 🤝 Contributing
Contributions are welcome! Feel free to open issues or submit pull requests.
📄 License
MIT License
⭐ Acknowledgements LangChain Chroma DashScope (Alibaba Cloud) 📌 Final Note
This project serves as a reusable AI agent foundation for building next-generation intelligent systems.