AntSK is a .NET-based AI knowledge base and intelligent agent platform designed for enterprise knowledge management, customer service, and intelligent applications. It empowers users to build custom GPT models, integrate external services via plugins, and leverage local or online AI models, including support for domestic models and offline operation.

How It Works

AntSK integrates Semantic Kernel for advanced natural language understanding and Kernel Memory for long-term knowledge storage and continuous learning. It supports document ingestion (Word, PDF, Excel, etc.) for Q&A, allows custom GPT creation, and exposes functionality through API and .NET plugin systems. Online search capabilities and model management for various providers (including llamafactory and ollama for local models) are core features.

Quick Start & Requirements

• Installation: docker-compose up -d (using provided docker-compose.yml and appsettings.json).
• Prerequisites: Docker, Docker Compose. Local model support requires Python and pip if not using pre-built images.
• Configuration: Requires appsettings.json for database connection (Postgres, Sqlite supported) and model paths.
• Demo: Available at the root of the repository, with default credentials test/test. Note: Demo server has limited permissions and cannot run local models.
• Documentation: README

Highlighted Details

• Supports local, offline AI model execution via llamafactory and ollama.
• Extensible via API and .NET DLL plugin systems.
• Handles document import from various formats (Word, PDF, Excel, PPT, etc.).
• Includes online search capabilities for real-time information access.

Maintenance & Community

• Active development indicated by recent updates.
• Contact via WeChat (xuzeyu91) or email (antskpro@qq.com) for support and discussion groups.

Licensing & Compatibility

• AntSK License, with additional terms based on Apache 2.0.
• Free Commercial Use: Allowed without code modification.
• Commercial License Required: For code modification, multi-tenant services (10+ users), bundling with hardware, or large-scale government/educational procurement.

Limitations & Caveats

• The online demo has restricted permissions and cannot run local models.
• Commercial use restrictions apply if specific conditions (code modification, multi-tenancy, bundling) are met, requiring a separate license.