RAGLight
RAGLight is a modular framework for Retrieval-Augmented Generation (RAG). It makes it easy to plug in different LLMs, embeddings, and vector stores, and now includes seamless MCP integration to connect external tools and data sources.
What is RAGLight?
RAGLight is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to raglight is a modular framework for retrieval-augmented generation (rag). it makes it easy to plug in different llms, embeddings, and vector stores, and now includes seamless mcp integration to connec...
RAGLight is a modular framework for Retrieval-Augmented Generation (RAG). It makes it easy to plug in different LLMs, embeddings, and vector stores, and now includes seamless MCP integration to connect external tools and data sources.
This server falls under the Data Science & ML category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- RAGLight is a modular framework for Retrieval-Augmented Gene
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx raglightConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use RAGLight
RAGLight is a modular Python framework for building Retrieval-Augmented Generation (RAG) pipelines that plugs into any combination of LLM provider, embedding model, and vector store without lock-in. It supports Ollama, LMStudio, Mistral, OpenAI, Gemini, AWS Bedrock, and vLLM for inference; HuggingFace or provider-native models for embeddings; and ChromaDB or Qdrant as vector stores. Through its MCP integration, RAGLight can also connect to external MCP servers as tool sources, enabling agentic RAG workflows that go beyond simple retrieval.
Prerequisites
- Python 3.10 or higher with pip
- At least one LLM provider accessible: Ollama running locally, or an API key for OPENAI_API_KEY, MISTRAL_API_KEY, GEMINI_API_KEY, or AWS Bedrock credentials
- Optional: a running vector store instance (ChromaDB or Qdrant) for persistent embeddings
- An MCP client such as Claude Desktop if using the MCP integration features
Install RAGLight
Install the base package with pip. Add optional extras for your preferred vector store or observability backend.
# Base install
pip install raglight
# With Qdrant (Windows-friendly, pure Python)
pip install "raglight[qdrant]"
# With ChromaDB
pip install "raglight[chroma]"
# With Langfuse observability
pip install "raglight[qdrant,langfuse]"Set your provider environment variables
Export credentials for whichever LLM and embedding providers you plan to use. RAGLight reads these at runtime.
# OpenAI
export OPENAI_API_KEY=sk-...
# Mistral
export MISTRAL_API_KEY=...
# Google Gemini
export GEMINI_API_KEY=...
# AWS Bedrock
export AWS_ACCESS_KEY_ID=...
export AWS_SECRET_ACCESS_KEY=...
export AWS_DEFAULT_REGION=us-east-1
# Ollama (local, no key needed)
export OLLAMA_CLIENT_URL=http://localhost:11434Build a basic RAG pipeline
Create a RAGPipeline with your chosen LLM and vector store config, build it (which indexes documents), then generate answers.
from raglight.rag.simple_rag_api import RAGPipeline
from raglight.config.rag_config import RAGConfig
from raglight.config.vector_store_config import VectorStoreConfig
from raglight.config.settings import Settings
config = RAGConfig(
llm=Settings.DEFAULT_LLM,
provider=Settings.OLLAMA
)
vs_config = VectorStoreConfig(
embedding_model=Settings.DEFAULT_EMBEDDINGS_MODEL,
database=Settings.CHROMA,
persist_directory='./raglight_db'
)
pipeline = RAGPipeline(config, vs_config)
pipeline.build()
response = pipeline.generate("What is the main topic of the documents?")
print(response)Start the REST API server
RAGLight ships with a built-in REST API and optional UI. Launch it with raglight serve and configure it via environment variables.
RAGLIGHT_LLM_MODEL=mistral-small-latest \
RAGLIGHT_LLM_PROVIDER=Mistral \
RAGLIGHT_DB=Qdrant \
raglight serve --port 8000 --uiEnable MCP server integration for agentic RAG
Pass an mcp_config list to AgenticRAGConfig to connect your RAG pipeline to external MCP servers as tool sources, enabling code execution, database access, or any other MCP tool alongside retrieval.
from raglight.rag.simple_agentic_rag_api import AgenticRAGPipeline
from raglight.config.agentic_rag_config import AgenticRAGConfig
config = AgenticRAGConfig(
provider=Settings.OPENAI,
model="gpt-4o",
mcp_config=[
{"url": "http://127.0.0.1:8001/sse"}
]
)
pipeline = AgenticRAGPipeline(config, vs_config)
pipeline.build()
response = pipeline.generate("Search the knowledge base and run the relevant code example")
print(response)RAGLight Examples
Client configuration
Claude Desktop configuration to connect to a RAGLight MCP server endpoint.
{
"mcpServers": {
"raglight": {
"command": "npx",
"args": ["raglight"],
"env": {
"OPENAI_API_KEY": "sk-your-key-here",
"RAGLIGHT_LLM_PROVIDER": "OpenAI",
"RAGLIGHT_LLM_MODEL": "gpt-4o",
"RAGLIGHT_DB": "Chroma"
}
}
}
}Prompts to try
Prompts for querying documents, ingesting new content, and testing agentic RAG capabilities.
- "What does the knowledge base say about the project architecture?"
- "Ingest all documents from the ./docs folder into the vector store"
- "Answer this question using hybrid search across all collections: what are the main API endpoints?"
- "Summarize the last 5 documents added to the knowledge base"
- "Run an agentic search and write a report on the findings from the knowledge base"Troubleshooting RAGLight
ChromaDB installation fails on Windows with C++ compiler error
Use the Qdrant backend instead: pip install "raglight[qdrant]". Qdrant is pure Python and has no native compilation requirements. Alternatively, install the Visual C++ Build Tools on Windows before installing chromadb.
RAG pipeline returns empty or irrelevant answers
Ensure documents have been ingested before calling generate(). Call pipeline.build() after providing the document source. Increase RAGLIGHT_K to retrieve more context chunks, and check that the embedding model matches the one used during ingestion.
MCP server URL in mcp_config is not reachable
Confirm the MCP server is running and accessible at the specified SSE URL before starting the agentic pipeline. Test connectivity with curl http://127.0.0.1:8001/sse. Ensure no firewall or port binding issue is preventing the connection.
Frequently Asked Questions about RAGLight
What is RAGLight?
RAGLight is a Model Context Protocol (MCP) server that raglight is a modular framework for retrieval-augmented generation (rag). it makes it easy to plug in different llms, embeddings, and vector stores, and now includes seamless mcp integration to connect external tools and data sources. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install RAGLight?
Follow the installation instructions on the RAGLight GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with RAGLight?
RAGLight works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is RAGLight free to use?
Yes, RAGLight is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
RAGLight Alternatives — Similar Data Science & ML Servers
Looking for alternatives to RAGLight? Here are other popular data science & ml servers you can use with Claude, Cursor, and VS Code.
Ultrarag
★ 5.6kA Low-Code MCP Framework for Building Complex and Innovative RAG Pipelines
RocketRide
★ 3.1k📇 🏠 - MCP server that exposes RocketRide AI pipelines as t
Aix Db
★ 2.1kAix-DB 基于 LangChain/LangGraph 框架，结合 MCP Skills 多智能体协作架构，实现自然语言到数据洞察的端到端转换。
NeMo Data Designer
★ 1.9k🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.
PaperBanana
★ 1.7kOpen source implementation and extension of Google Research’s PaperBanana for automated academic figures, diagrams, and research visuals, expanded to new domains like slide generation.
MiniMax
★ 1.5kBridges MiniMax AI capabilities to the Model Context Protocol, enabling AI agents to perform image understanding, text-to-image generation, and speech synthesis. It provides a standardized interface for accessing MiniMax's core tools via JSON-RPC.
Browse More Data Science & ML MCP Servers
Explore all data science & ml servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up RAGLight in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use RAGLight?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.