manthan89-py/Multi-Agent-Query-Resolution — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Python API for an AI customer support assistant for InfinitePay.

I want users to send a chat message and user ID, then the system should figure out whether the question is about InfinitePay products or a customer support issue. Product questions should search InfinitePay knowledge and answer accurately. Support questions should use customer context and past interactions, and create a support ticket when needed.

Please make the final answer friendly and easy to understand, but also return the raw agent answer and a simple trace showing which agent handled it and what tools were used. Add an endpoint to load the knowledge database before using chat, and make the API easy to test in Swagger at localhost.

Use Python with FastAPI, a multi agent setup, vector search for docs, and Mistral plus Tavily for AI and search. Include env setup for the API keys, Docker or Make commands to run it, and a few sample responses or tests so I can confirm the routing works.

Want more depth? Deep Reverse