Hey PH 👋 I built RAGPipe because I was tired of writing 40-line setup scripts every time I needed to add RAG to a project. LangChain is powerful but overkill for 90% of use cases. LlamaIndex is cleaner but still framework-y. What I wanted was the docker-compose of RAG — point it at data, it handles the rest, and it stays out of your way. So RAGPipe does one thing well: Sources → Transforms → Sinks. Files, git repos, or web pages in. Qdrant, Pinecone, or a JSON file out. Everything in between (chunking, embedding, cleaning) is automatic. The thing I'm most proud of is the CLI. `ragpipe watch .` auto-reindexes on file changes. `ragpipe git hook .` auto-indexes on every commit. `ragpipe serve` spins up a local API server your IDE or any tool can hit. Indexed the entire LangChain codebase (7,388 chunks) in 0.71s. No tricks. Would love feedback on: → What sources/sinks you'd want next → Whether the 3-function API feels right or too simple → Any edge cases in your data you've hit with other RAG tools pip install ragpipe-ai — and drop a ⭐ on GitHub if it saves you some boilerplate.

RAGPipe (OpenSource)

RAG in 3 lines. Zero config. Any data source.

RAG in 3 lines. Zero config. Any data source.