Agents Course

Click here to get the course: Agents Course A comprehensive course on building AI agents using Python, OpenAI, and Weave. This course covers various aspects of agent development, from basic workflows to complex multi-agent systems with memory and evaluation capabilities.

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/agents-course.git
   cd agents-course

2. Create and activate a virtual environment with all dependencies:

   Option A: Using uv (Recommended)

   uv venv .venv
   uv sync

   Option B: Using pip and requirements.txt

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

   Option C: Using pyenv and pip

   # Install Python 3.11 if not already installed
   pyenv install 3.11.0
   pyenv local 3.11.0
   
   # Create virtual environment
   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

   Option D: Using conda

   conda create -n agents-course python=3.11
   conda activate agents-course
   pip install -r requirements.txt

3. Set up your environment variables: Create a .env file in the project root with:

   WANDB_BASE_URL= # please set this if you are using dedicated cloud or onpremise
   OPENAI_API_KEY=your_openai_api_key
   WANDB_API_KEY=your_wandb_api_key

   To load these environment variables in your shell, you can run:

   set -a
   source .env
   set +a

Troubleshooting

ZoneInfoNotFoundError: 'No time zone found with key UTC'

If you encounter this error when running the evaluation scripts, try these solutions:

1. Ensure tzdata is installed (should be automatic with requirements.txt):

   pip install tzdata

2. For Windows users, you may need to explicitly install timezone data:

   pip install tzdata zoneinfo-backport

3. For Python < 3.9 users, install the backport:

   pip install backports.zoneinfo tzdata

4. Alternative workaround - Set timezone environment variable before running:

   # Linux/Mac
   export TZ=UTC
   python _5.2_evals.py
   
   # Windows (Command Prompt)
   set TZ=UTC
   python _5.2_evals.py
   
   # Windows (PowerShell)
   $env:TZ="UTC"
   python _5.2_evals.py

Course Structure & Order

The course is designed to be followed in order, with each module building on the previous one:

1. Basic Workflow (_1_workflow.py)
   Learn how to use OpenAI and Weave for simple prompt-response workflows.
   Run with:

   python _1_workflow.py

2. Simple Agent (_2_agent.py)
   Implement a minimal agent that can use tools and process user input.
   Run with:

   python _2_agent.py

3. Memory & Retrieval (_3_memory_retrieval.py)
   Add memory storage and retrieval capabilities to your agent, enabling it to remember and recall information.
   Run with:

   python _3_memory_retrieval.py

4. Multi-Agent Systems (_4_multi_agents.py)
   Build systems with multiple specialized agents that can hand off tasks to each other.
   Run with:

   python _4_multi_agents.py

5. Evaluation (_5_evals.py)
   Test and evaluate agent performance using automated evaluation tools and scenarios.
   Run with:

   python _5_evals.py

6. Simple Evaluations (_5_simple_evals.py)
   Quickly test agent responses with lightweight, script-based evaluation.
   Run with:

   python _5_simple_evals.py

7. MCP Integration (_6_mcp.py)
   Integrate with the Model Context Protocol (MCP) to allow agents to interact with the filesystem and external tools.
   Run with:

   python _6_mcp.py

Tip: Follow the modules in order for the best learning experience. Each script can be run independently as shown above.

Prerequisites

• Python 3.11 or higher
• uv package manager (recommended for dependency management)
• OpenAI API key
• Weights & Biases (wandb) API key
• npx (for MCP integration, see _6_mcp.py)

---

For more information or to get the course, visit: Agents Course