Team Mistral

src/README.md

@@ -7,118 +7,139 @@ This repository contains all materials for the hackathon focused on building age
 There are 2 pre-built MCP servers and a Jupyter notebook that can be run to execute LangChain to communicate with these MCP servers.
 
 - [Attractions MCP Server](mcp/attractions-mcp/README.md)
-- [Weather MCP Server](mcp/weather-mcp/README.md)
+- [Weather MCP Server (port 8009)](mcp/weather-mcp/README.md)
 - [Attractions Jupyter Notebook](agent/README.md)
+- [Macros MCP Server (port 8010)](mcp/macros-mcp/README.md)
+- [Nutritional Plan MCP Server (port 8011)](mcp/nutritional-plan-mcp/README.md)
+- [Trivia MCP Server (port 8012)](mcp/trivia-mcp/README.md)
 
 ## Getting Started
 
 Follow these steps to set up and run Weather MCP/Attractions MCP service:
 
 ### Prerequisites
+
 1. **Install Python 3.13**
+
    - Download from [python.org](https://www.python.org/downloads/)
 
 2. **Install uv** (Python package manager)
+
    ```bash
    pip install uv
    ```
-   
+
    **Windows users**: You may need to add the Python Scripts folder to your PATH environment variable:
+
    - If installed globally: `C:\Users\<your-username>\AppData\Local\Programs\Python\Python313\Scripts\`
    - If installed with `pip install --user uv`: `C:\Users\<your-username>\AppData\Roaming\Python\Python313\Scripts\`
    - Restart your command prompt/PowerShell after updating PATH
 
 ### Setup Steps
+
 1. **Fork and clone the repository**
+
    ```bash
    git clone <your-repo-url>
    cd Hackathon-2025
    ```
 
 2. **Navigate to the weather-mcp or attractions-mcp directory**
+
    ```bash
    cd src/mcp/weather-mcp
    ```
+
    ```bash
    cd src/mcp/attractions-mcp
    ```
 
 3. **Install dependencies**
+
    ```bash
    uv sync
    ```
 
 4. **Run the MCP server**
+
    ```bash
    uv run mcp dev main.py
    ```
+
    The weather MCP service will start and a local UI will launch in the browser (MCP's have to be running to run the jupyter notebook).
 
    When using mcp dev you may need to change the support type dropdown to STDIO to use it.
 
 > **⚠ Important Note:** If you are running the MCP to be consumed by the agent, do not run it in dev mode. Use the following command instead:
+>
 > ```bash
 > uv run main.py
 > ```
 
 ## Agent Notebook
+
 Once your MCP services are running, you can explore the agent implementation in the Jupyter notebook:
+
 - [Attractions Agent Notebook](agent/README.md)
 
 ## Creating a New MCP Service
 
 To create your own MCP service, follow these steps:
 
 1. **Navigate to the MCP folder**
+
    ```bash
    cd src/mcp
    ```
 
 2. **Initialize a new MCP project**
+
    ```bash
    uv init mymcpname-mcp
    ```
 
 3. **Add MCP dependencies**
+
    ```bash
    cd mymcpname-mcp
    uv add "mcp[cli]"
    ```
 
 4. **Create your MCP service code**
    Update the `main.py` file with this default MCP code to get you started:
-   
-   ```python   
+
+   ```python
    from mcp.server.fastmcp import FastMCP
-   
+
    # Create an MCP server
    # you can add the port here so that it doesnt clash with other mcp servers
    mcp = FastMCP("myname")
-   
+
    # Add an addition tool
    @mcp.tool()
    def add(a: int, b: int) -> int:
        """Add two numbers"""
        return a + b
-   
-   
+
+
    # Add a dynamic greeting resource
    @mcp.resource("greeting://{name}")
    def get_greeting(name: str) -> str:
        """Get a personalized greeting"""
        return f"Hello, {name}!"
 
-       
+
    if __name__ == "__main__":
       mcp.run("streamable-http")
    ```
 
 5. **Run your MCP service**
+
    ```bash
    uv run mcp dev main.py
    ```
 
 6. **Start building your custom functionality**
    - Follow the existing patterns in `weather-mcp` or `attractions-mcp` folders
    - Implement your custom tools and functionality
-   - for any additional guidance you can check here https://github.com/modelcontextprotocol/python-sdk
+   - for any additional guidance you can check here https://github.com/modelcontextprotocol/python-sdk

src/agent/README.md

@@ -18,6 +18,7 @@ This notebook demonstrates a LangChain agent integrated with MCP (Model Context
 ## Setup Instructions
 
 ### 1. Environment Setup
+
 Create a `.env` file in this directory with your Azure OpenAI credentials and mcp configuration:
 
 ```env
@@ -27,24 +28,29 @@ AZURE_API_VERSION=2024-12-01-preview
 DEPLOYMENT_NAME=
 ATTRACTIONS_MCP_URL=http://127.0.0.1:8008/mcp/
 WEATHER_MCP_URL=http://127.0.0.1:8009/mcp/
+MACROS_MCP_URL=http://127.0.0.1:8010/mcp/
+NUTRITION_PLAN_MCP_URL=http://127.0.0.1:8011/mcp/
 ```
 
 You can get these credentials from:
+
 - Azure OpenAI service → Azure AI Foundry Portal
 - Your Azure OpenAI resource in the Azure portal
 - Overview tab for endpoint, API key (selecting "Azure AI Services" tab within "Libraries")
 
 ### 2. Register Jupyter Kernel (Important!)
+
 Make sure you have downloaded the offical [Jupyter VS Code extension](https://marketplace.visualstudio.com/items?itemName=ms-toolsai.jupyter).
 
-**Important:** After opening the notebook file (*.ipynb), make sure to select the correct kernel:
+**Important:** After opening the notebook file (\*.ipynb), make sure to select the correct kernel:
+
 1. Install the offcial jupyter extension
 1. After opening the notebook file, click on "Select Kernel" and then "Python Environments...":
    ![alt text](images/image.png)
-2. Select "Create Python Environment"
-3. Select "venv` → Choose lastest python"
-4. Select the one `src\agent\requirements.txt` dependencies to install. Select ok on the prompt.
-4. This ensures you're using the local virtual environment, not a global one
+1. Select "Create Python Environment"
+1. Select "venv` → Choose lastest python"
+1. Select the one `src\agent\requirements.txt` dependencies to install. Select ok on the prompt.
+1. This ensures you're using the local virtual environment, not a global one
 
 This will create a virtual environment and start installing dependencies.
 
@@ -59,6 +65,7 @@ cd ../../mcp/attractions-mcp
 # Start the MCP server on HTTP
 uv run main.py
 ```
+
 The server should start on `http://localhost:8008/mcp/` for attractions
 
 ```bash
@@ -68,13 +75,15 @@ cd ../../mcp/weather-mcp
 # Start the MCP server on HTTP
 uv run main.py
 ```
+
 The server should start on `http://localhost:8009/mcp/` for weather
 
 ### 4. Running the Notebook
 
 Open and run the Jupyter notebook:
 
 You can verify the correct environment by running Cell 0 - it should show:
+
 ```
 python: *CodeDirectory*\Hackathon\Hackathon-2025\.venv\Scripts\python.exe
 uv: *Python install location*\Python313\Scripts\uv.EXE
@@ -91,33 +100,43 @@ You can hit `run all` at the top of the notebook, or you can run the cells 1 by
 The notebook is structured in sequential cells that should be run in order:
 
 ### Step 1: Environment Check
+
 - **Cell 0**: Verifies that the virtual environment is being used correctly
 
-### Step 2: Install Dependencies  
+### Step 2: Install Dependencies
+
 - **Cell 1**: Installs uv and all required packages from requirements.txt
 
 ### Step 3: Import and Setup
+
 - **Cell 2**: Imports all necessary libraries for LangChain and MCP integration
 
 ### Step 4-5: MCP Client Configuration
+
 - **Cell 3-4**: Sets up the MCP client to connect to the HTTP server on localhost:8008
 
 ### Step 6: Agent Initialization
+
 - **Cell 5-6**: Creates and initializes the LangChain agent with MCP tools
 
 ### Step 7: User Input Handler
+
 - **Cell 7**: Sets up functions to interact with the AI agent
 
 ### Step 8: Connection Testing
+
 - **Cell 8**: Tests the MCP server connection and displays available tools
 
 ### Step 9: Example Usage
+
 - **Cell 9**: Demonstrates how to use the agent with example queries
 
 ### Step 10: Interactive Chat
+
 - **Cell 10**: Interactive chat loop to converse with the agent
 
 ### Step 11: Cleanup
+
 - **Cell 11**: Provides cleanup function for proper resource management
 
 ## Available MCP Tools
@@ -140,11 +159,71 @@ Once the notebook is running, you can ask the assistant questions like:
 - "What's a random famous attraction I should visit?"
 - "Show me museums in London"
 
+## FastAPI Server (for Frontend Integration)
+
+If you want to expose the agent over HTTP so the React app can call it, run the FastAPI server that lives in `src/agent/server`.
+
+### 1. Configure environment
+
+Create or update your `.env` (or export variables) with either Azure OpenAI or OpenAI credentials:
+
+```bash
+# Azure (preferred)
+AZURE_OPENAI_ENDPOINT=
+AZURE_OPENAI_API_KEY=
+AZURE_API_VERSION=2024-12-01-preview
+DEPLOYMENT_NAME=
+
+# Optional fallback
+OPENAI_API_KEY=
+```
+
+Ensure the MCP servers you need (attractions, weather, macros, nutritional-plan, trivia) are running; the server will use their HTTP endpoints.
+
+Default URLs (override in `.env` if needed):
+
+```
+ATTRACTIONS_MCP_URL=http://127.0.0.1:8008/mcp/
+WEATHER_MCP_URL=http://127.0.0.1:8009/mcp/
+MACROS_MCP_URL=http://127.0.0.1:8010/mcp/
+NUTRITION_PLAN_MCP_URL=http://127.0.0.1:8011/mcp/
+TRIVIA_MCP_URL=http://127.0.0.1:8012/mcp/
+```
+
+### 2. Install dependencies & run server
+
+```bash
+cd src/agent
+python -m pip install -r requirements.txt
+python server/main.py
+```
+
+The server listens on `http://localhost:8080` by default and exposes:
+
+- `GET /health`
+- `POST /init` → `{ sessionId }`
+- `POST /chat` with `{ sessionId?, message }` → `{ sessionId, output }`
+
+Each session keeps in-memory chat history. Restarting the server clears sessions.
+Errors from MCP tools are returned to the caller as friendly messages in the API response frame.
+
+### 3. Frontend config
+
+Set the frontend environment variable so it knows where to reach the server:
+
+```
+src/frontend/.env
+VITE_AGENT_URL=http://localhost:8080
+```
+
+Then run the frontend (`npm run dev`) and use the chat UI.
+
 ## Troubleshooting
 
 ### Common Issues:
 
 1. **MCP Server Connection Failed**
+
    - Ensure the MCP server is running on localhost:8008
    - Check that no firewall is blocking the connection
 
@@ -153,9 +232,11 @@ Once the notebook is running, you can ask the assistant questions like:
    - Check that your Azure OpenAI deployment name matches (default: "gpt-5-chat")
 
 ### Debug Mode
+
 The agent runs in verbose mode by default, showing the reasoning process. To see detailed logs, check the console output when running cells.
 
 ## Project Structure
+
 ```
 src/agent
 ├── attractions.ipynb    # Main notebook file
@@ -164,8 +245,9 @@ src/agent
 ```
 
 ## Next Steps
+
 After setting up the notebook:
 
 1. Run through all cells sequentially
-3. Try your own travel-related questions
-4. Explore different MCP tools and their capabilities
+2. Try your own travel-related questions
+3. Explore different MCP tools and their capabilities