g3182479125-hub/assistgen-agent

AssistGen Agent is a self-hosted AI customer service system designed to be run locally on your own computer or server. It is written primarily in Chinese and targets developers or businesses who want to add an AI chat widget to an e-commerce website without sending data to a third-party service. The project includes both a backend server and a pre-built frontend that are packaged together. The backend is built with FastAPI, a Python web framework, and connects to AI language models through a standard API that works with several Chinese providers such as Kimi and DeepSeek. Users interact with the system through a chat interface that can be embedded on a shop page. Conversations are stored in a MySQL database, and the system supports user accounts with login and registration. There is also support for uploading documents and images so the AI can reference them during a conversation. The system has several optional components that can be added depending on your needs. Redis can be turned on to cache repeated questions so the AI does not have to be called again for the same query. Ollama allows you to run a local AI model for generating text similarity scores without making external API calls. Neo4j, a graph database, enables more complex relationship queries. GraphRAG is an optional knowledge-base feature that indexes documents and relationships together for more detailed question-answering. Setup involves creating a Python environment, editing a configuration file with your API keys and database password, creating the MySQL database, and starting the server. The README is in Chinese and includes step-by-step instructions for Windows using PowerShell. The repository only contains code and example configuration files, actual API keys, uploaded files, and logs are excluded and must be set up locally.