zilliztech/deep-searcher ? reverse-engineered prompt

Build me a Python app that can act like a deep research assistant for my private documents.

I want to drop in local files, maybe a whole folder, and ask questions in plain English, then get back a solid answer or a full report with sources pulled from that content. It should support common AI models like OpenAI, DeepSeek, Claude, Qwen, Grok, and Llama, and use a vector database like Milvus or Zilliz Cloud to search the data well. If it helps accuracy, it can also pull in web content, but the main focus should be private data.

Please make it easy to run locally, with clear setup, environment variable support for API keys, and a simple example I can try right away. If there are good defaults for embeddings and model settings, use them. Also make sure it’s organized enough that I can extend it later for enterprise knowledge search and Q and A.