A lightweight local code assistant built with Streamlit and managed with uv.
The goal is to load your codebase, embed it, turn it into RAG, and allow switching between multipl local LLMs (e.g., Llama, Qwen, Mistral). This tool gives you a private, local, developer-friendly environment for code search, explanation, refactoring suggestions, and project-aware answers.
-
Local Retrieval-Augmented Generation (RAG) over your entire codebase
-
Support for multiple models (Llama, Qwen, Mistral)
-
Simple Streamlit UI
-
uv-based environment + dependency management
-
Code chunking, embedding, and vector search
-
Local-only execution — no external API calls