AntSK

Overview of AntSK

What it is

AntSK is an enterprise-level AI knowledge base and agent platform built with .NET 9 and Blazor, integrating Semantic Kernel and Kernel Memory. It supports local offline AI large models, private deployment, and domestic (Chinese) technology stack adaptation. It can run completely without an internet connection and is designed to serve as the software base for an enterprise AI appliance (AI all-in-one machine). The project is hosted at https://github.com/shuyu-labs/AntSK and has a homepage at https://antsk.cn. It is categorized under AI on the Open Alternative platform and is open source with pricing type open_source. The license is stated as NOASSERTION, but the README specifies that the open source version is free for commercial use (requiring retention of logo and copyright info), while AntSK Pro is a commercial version.

Key features

• Semantic Kernel: natural language processing for understanding and responding to complex queries.
• Kernel Memory: long-term memory for storing knowledge points and personalizing interactions.
• Knowledge base: imports documents (Word, PDF, Excel, TXT, Markdown, JSON, PPT) and supports local BGE embedding and rerank models.
• Image generation: integrated with StableDiffusion local model for text-to-image.
• GPTs generation: platform allows creating personalized GPT models.
• API publishing: exposes internal functions as APIs for integration into other applications.
• API plugin system: open plugin system for third-party services.
• .NET plugin system: open DLL plugin system for business functions.
• Web search: real-time online search for up-to-date information.
• Model management: adapts and manages different models from various vendors, including local offline models via llama.cpp (GGUF), llamafactory, and ollama.
• Domestic technology support: supports domestic AI models and domestic databases (including Dameng), deployable in domestic technology environments.

Use cases

AntSK is designed for enterprise knowledge base applications, intelligent agent platforms, and AI all-in-one appliance deployment. It is suitable for scenarios requiring data to remain within the intranet, such as finance, government, healthcare, and military sectors with sensitive data. It is also intended for domestic technology (Xinchuang) environments and for .NET / C# development teams needing a native enterprise AI solution. It supports RAG-based question answering on imported documents and can be deployed offline without internet access.

Getting started

Docker one-line command start: minimum 4 CPU cores / 8GB RAM / 50GB SSD. Recommended 8 cores / 16GB RAM with optional GPU acceleration; production environment 16 cores / 32GB RAM / RTX 3080 or above. For detailed commands and configuration, see the repository README.

License and pricing

The open source version is free for commercial use, requiring retention of logo and copyright information. AntSK Pro is a commercial version with additional advanced features (GraphRAG, Text2Sql, visual workflow orchestration, and professional technical support). The license field in the metadata is NOASSERTION, but the README details the terms.

Based on the project's public README and repository metadata.