nginx_agent

nginx_agent is a stylish, AI-powered CLI assistant that helps you generate, explain, edit, lint, and manage NGINX configuration files. It uses a local Mistral model (via Ollama) and DuckDuckGo search to provide documentation-aware, secure, and customizable configurations — fast.

This project uses the Mistral model specifically for generating and editing NGINX configurations with doc-based reasoning, enhanced by live searches from official nginx.org.

---

Features

• Interactive config generation for:

• Static Content Serving

• Reverse Proxy

• Load Balancing (Round Robin, Least Connections, IP Hash)

• SSL/TLS Termination

• FastCGI/SCGI/uWSGI Proxying

• WebSocket Proxying

• HTTP Caching

• API Gateway (with JWT or API Key support)

• Mail Proxy (IMAP, POP3, SMTP)

• Security Enhancements (rate limiting, CSP, etc.)

• Custom configuration prompts

• Context-aware answers and code generation using DuckDuckGo and BeautifulSoup to scrape only the meaningful <div id="content">

• Cached documentation lookup from multiple sources

• Formatted and commented config output using shell-style comments

• Linting using nginx -t

• Terminal UI built with Questionary and Rich

• Supports full edits to existing configs via chat or raw editing

• Session state stored in memory for interactive workflows

---

Installation

1. Install requirements

pip install -r requirements.txt
sudo apt install nginx        # Or use brew install nginx on macOS

2. Download Ollama and pull Mistral

# https://ollama.com/download
ollama pull mistral

3. Create .env file

OLLAMA_MODEL=mistral

---

🔬 Usage

Launch the tool:

python main.py

Follow prompts like:

3. Load Balancing
Enter comma-separated backend URLs: http://1.1.1.1, http://2.2.2.2
Choose method: Round Robin
Enable health checks? Yes

Example generated config:

# Define upstream
upstream backend_servers {
    server 1.1.1.1;
    server 2.2.2.2;
}

server {
    listen 80;
    location / {
        proxy_pass http://backend_servers;
        # Add headers and timeouts here...
    }
}

After generation you can:

• Explain the config
• Edit or rewrite via AI
• Save as nginx.conf
• Search nginx.org
• Lint with nginx -t

---

Docker Compose Support

Choose "From Docker Compose" to analyze your docker-compose.yml, for example:

services:
  web:
    image: nginx
    ports:
      - "8080:80"

  api:
    image: node:18
    ports:
      - "3000:3000"

And get a matching proxy config automatically.

---

Smart Prompting & Search

• Prompts generated with detailed context-aware options
• Search hint is extracted from the menu choice
• Real docs fetched from DuckDuckGo and nginx.org
• Uses cache in .cache/ folder for repeat performance

---

Dependencies

rich
questionary
dotenv
ollama
duckduckgo-search
beautifulsoup4
pyyaml
requests

---

Relevant Links

• Mistral for the base model
• Ollama for local inference
• nginx.org for authoritative documentation

---

Built as a side project to explore the synergy between AI agents and intelligent web crawling for practical developer tooling.