CrewAI Grocery Planner

CrewAI Grocery Planner is a Python-based AI grocery planning assistant built with CrewAI. It helps generate meal plans, shopping lists, budget-friendly grocery recommendations, and leftover-reduction ideas using a multi-agent workflow.

Features

• Generates structured meal plans with ingredients and difficulty levels
• Organizes groceries into store sections for easier shopping
• Estimates costs and helps stay within a budget
• Suggests money-saving swaps and practical shopping tips
• Recommends ways to reuse leftovers and reduce food waste
• Supports dietary restrictions and cooking skill levels
• Uses optional web search via Serper for recipe research
• Produces structured outputs using Pydantic models

How It Works

The project defines a set of specialized agents:

• Meal Planner & Recipe Researcher — searches for a suitable recipe and builds a meal plan
• Shopping Organizer — groups ingredients into shopping categories
• Budget Advisor — reviews the plan for cost efficiency and suggests savings
• Leftover Manager — identifies leftover ingredients and suggests follow-up recipes
• Report Compiler — combines all outputs into one comprehensive guide

The main workflow in planner.py runs these agents sequentially to produce a full grocery planning experience.

Project Structure

.
├── planner.py
├── leftover.py
├── requirements.txt
├── .env.example
└── config/
    └── tasks.yaml

Requirements

• Python 3.10+
• An LLM provider API key
• Optional: Serper API key for recipe research

Dependencies

Installed from requirements.txt:

• crewai
• crewai-tools
• langchain
• langchain-community
• langchain-openai
• litellm
• python-dotenv
• duckduckgo-search
• databricks-sdk

Setup

1. Clone the repository

git clone https://github.com/xJoeyCodes/crewai-team.git
cd crewai-team

2. Create a virtual environment

python -m venv .venv
source .venv/bin/activate

On Windows:

python -m venv .venv
.venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Configure environment variables

Copy .env.example to .env and add your credentials:

cp .env.example .env

Example:

OPENAI_API_KEY=your-openai-key
LLM_MODEL=openai/gpt-4o-mini
SERPER_API_KEY=your-serper-key

You can also use other LiteLLM-compatible providers such as Anthropic or Groq by changing LLM_MODEL and setting the appropriate API key.

Usage

Run the planner:

python planner.py

By default, the script demonstrates a meal-planning workflow for:

• Meal: Chicken Stir Fry
• Servings: 4
• Budget: $25
• Dietary restrictions: no nuts, low sodium
• Cooking skill: beginner

Outputs

When you run the script, it can generate files such as:

• meals.json
• shopping_list.json
• shopping_guide.md

The workflow also prints structured example data to the console to show how the models are shaped.

Configuration

LLM model

planner.py reads the model from the LLM_MODEL environment variable:

llm_model = os.getenv("LLM_MODEL", "openai/gpt-4o-mini")

Optional search tool

If SERPER_API_KEY is set, the project uses SerperDevTool for recipe research.

Leftover Management

The LeftoversCrew in leftover.py loads task configuration from config/tasks.yaml and adds a leftover optimization step to the workflow.

This can help:

• identify ingredients likely to remain unused
• suggest extra recipes
• reduce food waste
• stretch the grocery budget further

Notes

• The current implementation is demo-oriented and runs immediately when planner.py is executed.
• The project is focused on a multi-agent grocery planning workflow rather than a reusable package.
• config/agents.yaml is referenced in leftover.py, so make sure the agent configuration is present if you extend that workflow.

Future Improvements

Possible next steps:

• add a CLI for custom meal input
• support multiple meal plans in one run
• add tests for the Pydantic models and workflow logic
• improve budget estimation and ingredient normalization
• expand YAML configuration for agents and tasks

License

No license file is currently included in this repository.