We've just released an update to RAGmate - an open-source RAG server that injects code context into JetBrains AI Assistant queries. It now supports storing and updating RAG indexes separately per Git branch.

This makes sure the context always matches the current code state without reindexing on every branch switch. The feature is implemented via Git hooks, and everything runs locally - no code is sent to the cloud.

RAGmate compatible with OpenAI, Ollama, HuggingFace, and other LLM APIs.

We welcome any feedback and ideas.