Atualizar README.md

README.md

@@ -33,3 +33,63 @@ Tudo isso dentro de uma **única conversa**, sem abas, com modos adaptativos:
 - cron jobs para monitoramento contínuo
 
 ---
+
+# Ollama Gateway – safira.renatorosa.com
+
+This project provides a simple HTTP gateway to a local Ollama instance,
+exposing multiple models (Llama 3.2, Gemma 4, and Qwen 2.5) through a
+Flask API served by Gunicorn inside Docker. The service is reachable
+over HTTP on port 80 at:
+
+**https://safira.renatorosa.com**
+
+Authentication is handled via Google OAuth.
+
+## Overview
+
+- **Backend:** Python (Flask) + Gunicorn  
+- **Models:** Llama 3.2, Gemma 4, Qwen 2.5  
+- **Hardware target:** Nvidia RTX 2070M GPU + 16 GB RAM  
+- **Auth:** Google OAuth 2.0  
+- **Runtime:** Docker + Docker Compose  
+- **Port:** 80 (HTTP)  
+
+Both the gateway and Ollama run inside the same `docker compose` file.
+
+## Environment variables
+
+Create a `.env` file:
+
+```
+GOOGLE_CLIENT_ID=your-google-client-id
+GOOGLE_CLIENT_SECRET=your-google-client-secret
+GOOGLE_REDIRECT_URI=https://.../oauth/callback
+FLASK_SECRET_KEY=change-this-secret
+OLLAMA_BASE_URL=http://ollama:11434
+```
+
+## Running
+
+docker compose up --build
+
+### Access:
+
+https://...
+
+## API
+
+### POST /api/generate
+
+```json
+{
+  "model": "llama3.2",
+  "prompt": "Explain quantum computing simply."
+}
+Response:
+
+json
+{
+  "model": "llama3.2",
+  "output": "..."
+}
+License