GH Repository · OpenBMB
UltraRAG
A Low-Code MCP Framework for Building Complex and Innovative RAG Pipelines
- stars
- 5,698
- 30-day movement
- +83/day
- Related entries
- 60
- Connections
- 3
python/uvPythonvlmhuggingface-transformersuieasydemopythonllmragqwenmultimodalembeddingmcpopenaivllmdeepseekdockergptflasksentence-transformers
UltraRAG is a low-code framework from OpenBMB for building RAG pipelines, structured around MCP (Model Context Protocol) servers so pipeline components can be composed without heavy coding. The repository ships a Flask-based UI and demos, and its topics indicate support for multiple LLM backends (OpenAI, vLLM, Qwen, DeepSeek) and multimodal/vision-language models.
You want to assemble complex or multimodal RAG pipelines visually or with minimal code instead of writing glue code yourself.
Use it to
- Compose RAG pipelines from MCP components
- Prototype multimodal retrieval with VLMs
- Swap LLM backends like vLLM, Qwen, or OpenAI
- Demo RAG applications through the built-in UI
- Provide pipeline context to coding agents via AGENTS.md/CLAUDE.md
For Developers and researchers building RAG applications
- Role
- rag
- Language
- Python
- Licence
- Apache-2.0
- Forks
- 449
- Open issues
- 4
- Last push
- 2026-09-16
- Latest release
- v0.2.0 · 2025-10-21
topicsragmcplow-codellmmultimodalpipelines