Recipe Creation & Evaluation System - Philer Onboarding Project

🏗️ Project Architecture

System Components

philer-onboarding/
│
├── agents/                     # Agents package - modular agent components
│   ├── __init__.py            # Package initialization and exports
│   ├── agent_definitions.py   # Agent class definitions and roles
│   ├── workflow_state.py      # Shared workflow state definition
│   ├── recipe_creator_node.py # Recipe creation node function
│   ├── recipe_evaluator_node.py # Recipe evaluation node function
│   └── format_output_node.py  # Output formatting node function
├── frontend/                   # Frontend web interface
│   ├── frontend.py            # FastAPI web server and chat interface
│   └── templates/             # Frontend HTML templates
│       └── chat.html          # Main chat interface
├── workflow.py                 # Main LangGraph workflow orchestration
├── main.py                     # Main entry point to run the frontend server
├── requirements.txt            # Python dependencies
├── .env.example               # Environment variables template
├── test_setup.py              # Setup validation and testing
├── setup.py                   # Quick setup automation script
└── README.md                  # This file

🤖 Agent System Design

Agent 1: Recipe Creator Agent

• Role: Creative Recipe Developer
• Purpose: Generate detailed, practical recipes based on user requests
• Capabilities:
  • Creates complete recipes with ingredients and instructions
  • Considers dietary restrictions and preferences
  • Provides cooking tips and variations
  • Estimates cooking and prep times

Agent 2: Recipe Evaluator Agent

• Role: Recipe Quality Assurance Specialist
• Purpose: Evaluate recipes for logic, safety, and quality
• Capabilities:
  • Analyzes cooking procedures for correctness
  • Identifies potential food safety issues
  • Suggests recipe improvements
  • Validates ingredient combinations and cooking methods

📁 File System Guide

agents.py - Agent Definitions

This file contains the core agent classes and their configurations:

# Key Components:
- RecipeCreatorAgent class: Handles recipe generation
- RecipeEvaluatorAgent class: Handles recipe evaluation
- AGENT_DEFINITIONS: Registry of all available agents

Role: Defines the "who" - what each agent does and how they behave

workflow.py - LangGraph Workflow

This is the main orchestration file that defines how agents work together:

# Key Components:
- WorkflowState: Defines data structure passed between agents
- RecipeWorkflow class: Orchestrates the multi-agent process
- Node functions: Individual steps in the workflow
- Graph building: Defines the flow between agents

Role: Defines the "how" - the sequence and logic of agent interactions

🚀 Setup Instructions

1. Environment Setup

# Create and activate virtual environment
python -m venv venv
venv\Scripts\activate  # Windows
source venv/bin/activate  # Linux/Mac

# Install Python dependencies
pip install -r requirements.txt

# Set up environment variables
copy .env.example .env
# Edit .env file and add your OpenAI API key

2. Configuration

Edit the .env file:

OPENAI_API_KEY=your_openai_api_key_here

3. Running the Application

Option A: Run the Full Web Application (Recommended)

python main.py

Then open your browser to: http://localhost:8000

🔄 Workflow Flow Explanation

Step-by-Step Process

1. User Input: User requests a recipe via the chat interface
2. Recipe Creation: Recipe Creator Agent generates a detailed recipe
3. Recipe Evaluation: Recipe Evaluator Agent reviews and critiques the recipe
4. Output Formatting: System combines both outputs into a user-friendly response
5. Response Delivery: Final result is displayed in the chat interface

State Management

The workflow uses a WorkflowState object that gets passed between nodes (this is easier for keeping track of progress and variables):

WorkflowState:
- user_input: Original user request
- recipe: Generated recipe
- evaluation: Professional evaluation
- final_output: Formatted final response
- step: Current workflow stage

🔧 Customization Ideas

1. Read the LangGraph documentation
2. Make custom agents (e.g. specializing in Chinese cuisine by editing agent role)
3. Add more agents (cost calculators)