Mintplex-Labs/anything-llm ? reverse-engineered prompt

I want to build an application that's like a private, self-hosted ChatGPT. The main idea is that it should be privacy-focused and run locally, so I can chat with my own documents—like PDFs, Word docs, and text files—without sending them to a third party. The chat interface should be clean and intuitive, with drag-and-drop for file uploads, and it needs to show citations so I know exactly where in the documents the answers are coming from.

It needs to be super flexible, allowing me to connect to lots of different LLMs. I want to easily switch between using a local model with Ollama or LM Studio and a cloud service like OpenAI, Anthropic, or Gemini. It should also be a multi-user application so I can set up accounts for my team. Please also include a feature for creating custom AI agents that can perform tasks, and a no-code builder for them would be great. The whole thing should be easy to set up and deploy using Docker.