CrewAI YouTube Workflow

This repository is a small CrewAI project that turns a topic into a short YouTube-style script.

It uses two agents:

• A researcher agent that gathers simple facts using practical tools.
• A script writer agent that turns that research into a short, engaging script.

The workflow is defined in main.py, with the agents in agents.py and the tasks in tasks.py. The helper tools live in tools.py.

What it does

When you run the project, it:

1. Takes a topic from the command line or falls back to a default topic.
2. Runs a research task first.
3. Runs a script-writing task second.
4. Prints the final result in the terminal.

The researcher agent currently has two practical tools:

• Wikipedia summary lookup
• DuckDuckGo instant answer lookup

Project files

• main.py: entry point that builds and runs the CrewAI crew.
• agents.py: defines the researcher and script writer agents.
• tasks.py: defines the research and script-writing tasks.
• tools.py: defines the tools used by the researcher agent.

Requirements

• Python 3.10+ recommended
• crewai
• litellm
• A Hugging Face inference token set in your environment

How to run

From the project folder:

./venv/Scripts/Activate.ps1
$env:HF_TOKEN="hf_your_token_here"
python main.py

You can also pass your own topic:

python main.py "What is agentic AI?"

Notes

• The repo is meant as a simple starting point for a multi-agent workflow.
• You can later expand it with more tools, more tasks, or a richer research pipeline.