Everything you need to actually run the labs: the build-only model, the exact folder layout (where your agents get created), one environment setup, a complete copy-paste prompt for every lab, and links to all the source docs.
Each module has one copy-paste prompt. You paste it into a fresh AI-coding-agent session; the agent builds the lab files but does not run them. You run each lab yourself in your terminal — that's how you see the colored traces and the interactive prompt.
my-agents/.OPENAI_API_KEY and run the lab yourself to see traces + interactive mode.Your coding agent builds every lab into my-agents/. The docs you paste from live in course/agent-docs/; reference implementations live in course/labs/.
Python 3.11+ required. Do this once; later modules reuse the same venv.
mkdir -p my-agents
python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install "deepagents>=0.4.3" \
"langchain-openai>=1.1.8" \
"openai>=1.66.0"
export OPENAI_API_KEY="sk-your-key-here"
mkdir my-agents
python -m venv .venv
.venv\Scripts\activate
pip install --upgrade pip
pip install "deepagents>=0.4.3" ^
"langchain-openai>=1.1.8" ^
"openai>=1.66.0"
set OPENAI_API_KEY=sk-your-key-here
OPENAI_API_KEY must start with "sk-" — get one at platform.openai.com/api-keysEach prompt is complete — it includes the venv setup, the exact pip install, trace_utils.py, and the full lab code. Click Copy, paste into a fresh Codex / Claude Code / Cursor session in an empty folder, and it builds the files (no repo needed). You then run the lab yourself.
The full markdown your agent consumes — open or download any of them.