OPENAI PATH · O3 · AGENTS API
Build with managed agent sessions
Understand the Agents API session lifecycle, events, recovery and cleanup.
45–60 minutes · Python 3.12+ · Notebook + terminal · Local core needs no API key.
Start with a testable local contract.
Download and extract the lab on your laptop. Open lesson.ipynb or follow README.md. Use Python 3.12+. The local core uses fictional fixtures and does not call an API.
Work inside the extracted 10-openai-managed-agents folder. The starter tests intentionally fail until you implement the TODO functions in exercise.py.
Notebook and Windows setup
Open lesson.ipynb using your existing Jupyter environment, or install JupyterLab in the virtual environment.
python -m pip install jupyterlab
python -m jupyterlab lesson.ipynbWindows PowerShell activation:
py -3.12 -m venv .venv
.venv\Scripts\Activate.ps1python3 -m venv .venv
source .venv/bin/activate
python -m unittest -v test_exerciseDone when The starter tests run and you can explain the first TODO.
Sign in to save progress and private notes. The app does not execute or grade your local code.
YOUR EVIDENCE
What actually happened?
Go deeper at the source.
These are original companion exercises, not official OpenAI training. Sources and SDK versions were checked 2026-10-04. The download includes precise source references and the distinction between fixture tests and real API behavior.
Agents API overviewSigned-in profiles and learning activity are visible to the app owner. Project notes stay private. Your profile & data