Full-day hands-on workshop — Revision 1.12 — 09/23/26
AI agents become useful when they can act. In this workshop you build a practical AI agent that uses tools through two different surfaces — two ways of offering a tool to an agent: command-line interfaces and the Model Context Protocol (MCP) — and learn when each one is the right engineering choice.
Over ten short labs (10–12 minutes each) you will:
- Build a Python agent that can reason over a task, choose a tool, execute it, and use the result
- Give the agent tools implemented as local CLI commands
- Design agent-friendly tools with clear inputs, predictable outputs, and meaningful errors
- Build an MCP server and connect your agent to it as an MCP client
- Compare the CLI and MCP approaches head-to-head on the same task
- Wrap an existing CLI (git) behind an MCP server
- Add guardrails: allowlists, parameter validation, human approval, logging, and evaluation checks
These instructions will guide you through configuring a GitHub Codespaces environment that you can use to run the course labs.
1. Change your codespace's default timeout from 30 minutes to longer.
To do this, when logged in to GitHub, go to https://github.com/settings/codespaces and scroll down on that page until you see the Default idle timeout section. Adjust the value as desired.
2. Click on the button below to start a new codespace from this repository.
Click here ➡️ 
3. Then click on the option to create a new codespace.
This will run for several minutes while it gets everything ready.
If VS Code shows a workspace trust prompt, click Trust Folder & Continue.
After the initial startup, it will run a script to set up the Python environment, install
Ollama, and download the llama3.2:3b model. This takes several more minutes. The
codespace is ready when the terminal shows Ollama ready with llama3.2:3b. and a prompt.
Verify with:
ollama list
4. (Recommended) Get a free API key for Groq to use a larger, faster model.
The labs run entirely on the local llama3.2:3b model by default, and everything works
that way. But local model calls take 5-10 seconds each on a codespace, and a small model
occasionally gives a muddled answer — part of what we discuss in the workshop. Groq hosts
a larger model for free (no credit card) that answers in about a second. Labs 7 and 10,
which run several agents back to back, are noticeably nicer with it.
a. In a browser, go to https://console.groq.com and create an account. (If you get an email with a confirmation button, make sure the link opens in the same browser you used for Groq. If not, copy the link from the "click here" section and paste it into the right browser.)
b. In the top right of the Groq screen, click on API Keys
c. Then click the Create API Key button.
d. Fill in the information, verify you're human if asked, and click Submit.
e. Copy the key (you can't view it again later).
5. Set up your Groq key in your codespace.
Back in the codespace TERMINAL, run the command below to set your key for this and all future terminals. Paste your key when prompted and hit Enter:
source scripts/setup-key.sh
You should see Done! GROQ_API_KEY is set .... Every lab program checks that variable:
set means Groq's qwen/qwen3.8-27b model, unset means the local Ollama model. The
setting persists across new terminals and codespace restarts.
To confirm the key works and the labs' model is reachable, run:
bash scripts/check-groq.sh
You should see OK labs model -> qwen/qwen3.8-27b. If it reports FAIL because
Groq has retired that model, the script checks replacements for you and prints the exact
export GROQ_MODEL=... line to run — or tells you to go back to the local model.
6. Open labs.md and start with Lab 1.
Right-click labs.md in the Explorer and choose Open Preview for the rendered version.
If you prefer to run locally instead of in a Codespace, you need:
- Docker Desktop and VS Code with the Dev Containers extension (open the repo folder and choose "Reopen in Container"), or
- Python 3.10+,
pip install -r requirements.txt, and Ollama installed from https://ollama.com withollama pull llama3.2:3b
Materials in this repository are for educational use only by attendees of our workshops.
(c) 2026 Tech Skills Transformations and Brent C. Laster. All rights reserved.