davidmigloz/langchain_dart

Provides unified abstractions across multiple LLM providers (OpenAI, Google, Anthropic, Ollama, etc.) and implements core RAG patterns including document loading, embedding, vector storage, and agentic tool use through composable LCEL chains. The modular architecture splits functionality into `langchain_core` (abstractions), `langchain` (high-level chains/agents), and provider-specific packages, allowing developers to depend only on required components.

668 stars. Actively maintained with 4 commits in the last 30 days.