yuan7code/meddra-coding-ai ? reverse-engineered prompt

Build me a Python app for MedDRA coding that helps a user paste in a free text medical term and get the most likely MedDRA code back, along with a clear explanation of why it matched.

I want it to read MedDRA ASCII dictionary folders, build searchable vector indexes for each language and version, and then use retrieval plus an LLM to pick the best term from the top matches. The interface should be a simple Streamlit page with a chat style input, a sidebar that shows the closest results and similarity scores, and a main area that shows the selected code and the full hierarchy from LLT up to SOC.

Please make it work with OpenAI, OpenRouter, or a local Ollama model, and let me configure paths, embeddings, and retrieval settings from a config file. Include a command line way to rebuild indexes for all versions or just one version. If you need to, look up the current docs online for the libraries you use.