# ⛅ Weather MCP Server

**Protocol:** MCP | **Python 3.11+** | **License:** MIT | **Status:** Active ✅

A robust project implementing the **Model Context Protocol (MCP)** architecture — an autonomous LLM-powered agent capable of fetching up-to-date weather forecasts for cities in Israel and the United States, providing seamless natural language responses in a terminal chat interface.

---

## Project Overview

This project extends LLM capabilities by granting access to external data sources that are not natively available: retrieving live forecasts from sites that lack an official public API. 

The system bridges this gap using two complementary retrieval engines:

| # | Information Source | Mechanism & Technical Description |
|---|---|---|
| **1** | **Israel Server (`weather-Israel`)** | **Browser Automation (Playwright):** Launches a headless browser instance navigating to `weather2day.co.il/forecast`, locates the target city, extracts the page content, and feeds it directly to the LLM for analysis (RAG approach). |
| **2** | **USA Server (`weather-USA`)** | **Official API REST Calls:** A lightweight server executing direct REST requests against the official National Weather Service (NWS) API. |

The **Host** layer orchestrates the flow: connecting servers to the language model, dynamically exposing available tools, and managing the conversation loop until the model yields an accurate response.

---

## Workflow Architecture

```text
User: "What's the weather in Tel Aviv?"
    │
    ▼
Host ◄──► LLM autonomously selects and executes tools
    │
    ▼
Israel Server (Sequential pipeline):
open ──► enter("Tel Aviv") ──► select ──► get_content
    │
    ▼
Content analysis (RAG) by LLM
    │
    ▼
✨ Final response generated in natural language