nanda050502/Zagent_001V2 — reverse-engineered prompt

Build me a production ready Python backend for company intelligence. I want to send basic company details like name, website, logo, industry, location, employee size, CEO, LinkedIn, and a few settings, then get back a complete structured company profile with exactly 165 fields.

Use LangGraph with FastAPI and connect it to Gemini, Groq, and OpenRouter. The system should research the company, generate results from the providers, retry when fields are missing or invalid, validate the response strictly, and then consolidate the best final answer. If providers disagree, pick the majority answer first, otherwise use the research context to decide.

Add simple API docs, a health endpoint, and a main endpoint for generating the profile. Store individual provider outputs and the final consolidated profile in Supabase when enabled. Make it configurable with environment variables for API keys, tracing, web search settings, and storage. Also include Docker support and a basic Jenkins pipeline that builds, runs, and smoke tests the service.