dingent 0.5.12b19

AI agent framework

This project is currently under rapid development. For the latest and most accurate documentation, please visit: ngdc.cncb.ac.cn/dingent

最新文档和使用指南，请访问： ngdc.cncb.ac.cn/dingent

Dingent

A powerful, yet user-friendly LLM Agent framework designed to streamline the entire development lifecycle of intelligent applications.

Dingent is an agent framework whose core goal is to simplify the process of creating any application powered by Large Language Models (LLMs). We provide a concise yet powerful toolkit... to build applications capable of automating complex workflows, interacting with various services, and performing intelligent analysis. For any custom logic or integration, Dingent offers a flexible framework that developers can easily extend by writing custom code.

Chat Interface

Chat View 1	Chat View 2

Admin Dashboard

Dashboard - Overview	Dashboard - Workflows
Dashboard - Settings	Dashboard - Logs
Dashboard - Assistants	Dashboard - Market

🎯 Why Choose Dingent?

When building LLM applications, developers often spend a significant amount of time on "glue code": creating backend services, wrapping APIs, setting up frontend-backend communication... These tasks are tedious and repetitive.

Dingent's core value lies in:

• No More Repetition: We package the best practices for backend services (LangGraph), data interfaces (Plugin System), a chat interface (CopilotKit), and a full-featured admin dashboard into a single command. You no longer need to build everything from scratch and can start writing your core business logic immediately.

• Configuration via UI: Forget manually editing complex configuration files. With Dingent's integrated admin dashboard, you can manage assistants, build workflows, and configure settings through an intuitive graphical interface.
• Extensible and Versatile: While Dingent began with a focus on data retrieval, it has evolved into a powerful general-purpose framework. Its modular architecture and robust plugin system allow you to build any type of agent—from simple task automation bots to complex multi-agent systems. Dingent provides the solid foundation, you bring the vision.
• Core Features Built-In: We believe a simple and easy-to-use agent shouldn't require users to spend a lot of time maintaining plugins. Therefore, we are committed to integrating features the community deems important directly into the framework. If you think a feature is crucial, we encourage you to open an Issue or PR. This directly reflects our core mission of "making Agents simpler for users."

• Smooth Learning Curve: You only need a basic understanding of Python to build powerful, general-purpose agents, without needing to be an expert in LangGraph or FastAPI. At the same time, we retain the flexibility to expand functionalities, ensuring the framework can fully support custom development when needed.

✨ Features

• Instant Project Setup: Simply run dingent dev in any directory to initialize a new project.
• Integrated Admin Dashboard: A powerful, React-based web interface to visually manage your assistants, workflows, plugins, and settings.
• Bundled Frontend: A pre-built, standalone Next.js chat interface is included out-of-the-box. No need for manual setup or compilation.
• Lightweight and Easy to Use: A clean design and a gentle learning curve let you focus on business logic rather than tedious configuration.
• Guest Mode Support: Allow unauthenticated users to chat with AI agents without requiring registration. See Guest Mode Documentation for details.

🚀 Quick Start

Just download the latest executable for your platform from the releases page and run it.

🗺️ Roadmap

• ✅ 1. Documentation & Tutorials

  • Basic Docs: Installation and configuration guides.
  • Core Concepts: In-depth explanations of key features and design.
  • Plugin Dev Guide: How to build your own plugins.
  • End-to-End Tutorials: Complete, step-by-step project examples.

• ✅ 2. Admin Dashboard

  • Core UI: Fully functional dashboard for managing assistants, workflows, and settings.
  • Plugin Management: Add/remove plugins directly from the UI.
  • Advanced Workflow Editor: Visual node-based tools for building complex agent behaviors and logic flow.

• ✅ 3. Plugin System

  • Auto-Discovery: Automatically loads plugins from the plugins/ directory.
  • Plugin CLI: Install and manage plugins via the command line. (Superseded by UI management in the Admin Dashboard)
  • Plugin Marketplace: Discover, search, and dynamically mount community plugins via the dashboard.

• ✅ 4. Core Plugins

  • Database Plugin: Connect to mainstream databases via specialized Text2SQL engines.
  • Knowledge Base Q&A Plugin: Intelligent retrieval and QA using entity-enhanced RAG and vector databases.
  • Heterogeneous Integration: Out-of-the-box support for cross-platform data sources (e.g., GenBase, iDog, BioKA).

• ✅ 5. Core Architecture & Deployment

  • Dynamic Workflow Engine: GraphFactory for zero-code, Just-In-Time (JIT) compilation of multi-agent state machines.
  • MCP Native: Deep integration with the Model Context Protocol for decoupled, decentralized tool execution and sandboxing.
  • Zero-Config Deployment: Heterogeneous dual-process packaging (Next.js + PyInstaller) for seamless cross-platform setup.

🤝 How to Contribute

We created this project to make Agents simpler for users, not to build yet another complex development framework. Therefore, we warmly welcome and heavily rely on community contributions to shape the future of Dingent.

If a feature is important to you, we strongly encourage you to discuss it by opening a GitHub Issue or contributing code directly through a Pull Request. Our core philosophy is that the developer community should decide which features are built into the software, rather than leaving users to maintain their own plugins!

If you share our vision and wish to contribute code, please follow these steps:

1. Fork this repository.
2. Create a new feature branch (git checkout -b feature/YourAmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push your branch to GitHub (git push origin feature/YourAmazingFeature).
5. Create a Pull Request and clearly describe the value of your feature.

We believe that through our collective efforts, Dingent can become a truly powerful and "out-of-the-box" tool.

📄 License

This project is licensed under the MIT License.