Show HN: Dart_agent_core – Run AI agents in Flutter apps with lifecycle hooks (github.com)

A new Dart library named **dart_agent_core** has been announced, designed specifically to facilitate the development and execution of AI agents within Flutter applications. This library enables the creation of stateful AI agents that leverage various tools using a simple interface, incorporating features like state persistence, multi-turn memory, and skill systems, while connecting seamlessly to popular LLM providers such as OpenAI, Google Gemini, and Anthropic Claude. Its mobile-first design allows developers to build sophisticated AI functionalities directly within Dart and Flutter without requiring backend support from Python or Node.js. The significance of dart_agent_core lies in its comprehensive orchestration capabilities for AI interactions. It allows developers to wrap Dart functions as tools, manage stateful sessions, and handle multimodal inputs, making it suitable for real-time, responsive applications. Noteworthy technical features include context compression for memory management, agent evaluation functionality, and support for dynamic skills that can be activated or deactivated based on the conversation context. Additionally, the library’s **runStream()** method enhances user experience by providing live updates for UI interaction during AI processing, making it a powerful tool for developers looking to integrate AI agents into their applications efficiently.