AIA26

Welcome to AIA26! This repo is your companion for the workshop. Follow the blocks in order, write some code, and leave with a working AI agent.

Getting Started

1. Create a GitHub Codespace

From previous workshops we know that the project setup does not always go smoothly. Therefore, we recommend creating a GitHub Codespace. Personal GitHub accounts offer 120h of free Codespace usage per month, which boils down to 60h of usage if you choose the smaller 2-core machine.

Click here to create your Codespace for this workshop. Alternatively, if you don't want to use GitHub Codespaces, you can clone this repository to your local machine and set up everything there.

Important

Do not forget to stop your Codespace after the workshop to save the remaining hours.

2. Install dependencies

This repo uses uv for dependency management. From the repository root:

uv sync

3. Configure your environment

Copy the template:

cp .env.template .env

4. Get your API keys and store them in .env

Across the exercises you'll need accounts / keys for:

• OpenRouter — https://openrouter.ai (OPENROUTER_API_KEY)
• Tavily — https://tavily.com (TAVILY_API_KEY)
• Langfuse Cloud — https://cloud.langfuse.com (LANGFUSE_SECRET_KEY, LANGFUSE_PUBLIC_KEY, LANGFUSE_BASE_URL)

OpenRouter setup

You should have received an OpenRouter API key from us. Please paste it into the .env file in the root of this repository.

Langfuse setup

1. Register on https://cloud.langfuse.com/.
2. Create an organization (e.g. "Personal").
3. Add a project (e.g. "AIA26").
4. In the project settings, click API Keys → Create new API keys and copy the .env block Langfuse shows you, then paste it into the .env file in the root of this repository.

Tavily setup

1. Register on https://app.tavily.com/home.
2. Under "API Keys", click on the + icon.
3. Click the copy icon to copy your new key and paste it into .env file in the root of this repository.

5. Install Deep Agents Code (dcode) for skill exercises

Some blocks use the skill-creator skill under .deepagents/skills/skill-creator/. Run it with Deep Agents Code:

uv tool install 'deepagents-code[openrouter]'

Add your OpenRouter key to dcode's config (in addition to the repo .env used by Python exercises):

mkdir -p ~/.deepagents
echo 'OPENROUTER_API_KEY=your-key-here' >> ~/.deepagents/.env

From the repository root:

dcode --default-model openrouter:openai/gpt-oss-120b

Then invoke the skill creator with /skill:skill-creator or:

dcode --skill skill-creator -n "Help me build a skill for …"

---

Well done! You're now ready to start with exercises/block1.