💡 Easy to Use
Simple API design, create intelligent agents with just a few lines of code, supporting rapid prototyping and production deployment
RWKV-based
Agent Development Framework
Providing AI agents with true memory and thinking capabilities
Simple API design, create intelligent agents with just a few lines of code, supporting rapid prototyping and production deployment
Long-term memory system based on vector databases, enabling AI agents to remember and learn from historical conversations
Efficient inference engine based on RWKV models, supporting CPU inference with millisecond-level response speed
Flexible tool system and plugin architecture, easily integrate various external services and functional modules
Native RWKV model support, intelligent state management, multimodal input, streaming output, optimized inference performance
Agent orchestration and collaboration, role customization, communication mechanisms, load balancing, intelligent task allocation
Dynamic short-term memory management, persistent long-term memory storage, semantic fragment aggregation, graph storage, personal profiling
Model2Vec integration, semantic search, hybrid retrieval, performance optimization, query optimizer and caching mechanisms
Tool registry, shared tools, extension interfaces, error handling, cross-agent tool sharing mechanisms
SQLite database, unified interface, data migration, performance monitoring, lightweight yet powerful local storage