travel-planner

# 📍 Travel Planner – AI Agent System This project is a multi-agent Travel Planner powered by the **A2A (Agent-to-Agent)** protocol. It uses LangChain, Ollama LLM, and multiple A2A-compatible agents to collaboratively fetch weather data, search results, and generate a final travel recommendation.  --- ### 🚀 Features * 🌦️ **WeatherAgent** – Provides weather info for the destination. * 🔍 **TavilySearchAgent** – Searches the web for relevant information. * 🤖 **LocalLLM** – Hosts a local LLM (`llama3.2`) using `langchain_ollama`. * 🧠 **TravelPlanner** – Coordinates the above agents using the A2A protocol. --- ### 🧩 Architecture ```mermaid graph TD; User --> TravelPlanner TravelPlanner --> WeatherAgent TravelPlanner --> TavilySearchAgent TravelPlanner --> LocalLLM ``` --- ### 📁 Directory Structure ``` travel-planner/ │ ├── .env # Environment variables ├── WeatherAgent.py # Weather API Agent (A2A) ├── TavilySearchAgent.py # Search Agent using Tavily (A2A) ├── LocalLLM.py # Local LLM server using Ollama + LangChain ├── Travel_Planner_Agent.py # Main Planner Agent ├── TravelPlannerApp.py # Main Streamlit App ├── main.py # Runs agents sequentially with delay └── README.md # This file ``` --- ### ⚙️ Setup #### 1. 📦 Install Dependencies ```bash uv venv .venv\Scripts\activate uv sync ``` #### 2. 🧪 Configure `.env` Ensure you have the following keys in your `.env` file: ```dotenv OPENWEATHER_API_KEY=your_weather_key TAVILY_API_KEY=your_tavily_key ``` #### 3. 🏃‍♂️ Run All Agents ```bash uv run main.py ``` This script launches all agents one by one with a 2-second delay. --- ### 🛠 Tech Stack * 🧠 `LangChain` + `Ollama` * 🔄 `python_a2a` (Agent-to-Agent protocol) * 🌐 `FastAPI` under the hood * ⛅ Weather & Search APIs --- ### 📝 Notes * All agents expose A2A-compatible APIs. * The servers **must run concurrently** for A2A to function properly. * Best used with `llama3.2` loaded in your Ollama setup.