RAGLight

Overview of RAGLight

What it is

RAGLight is a small, typed Python toolkit for composing classic and agentic retrieval-augmented generation (RAG) pipelines. The core package has no third-party runtime dependency; applications inject an LLM, an embedding model, a vector store, and optional behavior through public structural protocols. It includes seamless MCP integration to connect external tools and data sources.

Key features

• Modular design with structural protocols for custom components (no inheritance or registry required)
• Supports classic RAG pipelines and agentic RAG with tool use and MCP adapters - Stateless pipeline that can be shared between users; session management for isolated conversation history - Conversational query reformulation (opt-in) using the configured LLM - Directory ingestion with recursive traversal, chunking, and deterministic ordering - Built-in integrations for Ollama, OpenAI, Mistral, Gemini, Bedrock, Hugging Face, Chroma, Qdrant, Langfuse, and agentic MCP - Provider-neutral retrieval filters (And, Or, Eq, In, MatchNone)

Use cases

• Classic RAG with local or cloud LLMs and vector stores - Agentic RAG where the pipeline can call external tools (e.g., weather API) via MCP or custom function tools - Ingesting and indexing document directories with configurable chunking - Building custom components (LLM, embeddings, vector store, reranker) by implementing structural protocols - Multi-user applications with stateless pipeline sharing and isolated sessions

Getting started

bash pip install raglight pip install 'raglight[ollama,chroma,documents]' Available extras include openai, ollama, mistral, gemini, bedrock, huggingface, documents, chroma, qdrant, langfuse, and agentic-mcp.

License and pricing

Licensed under the MIT License. The project is open source and free to use.

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