Skip to main content

Examples

Examples are working definitions you can copy into your YAML editor and adapt to your own use case. They cover the common building blocks — running notebooks, creating AI Engines, and sending notifications. Each example links to its own page with the full YAML rendered inline.

The reusable examples (everything except the scheduled notebook) are composite actions — publish them once, then call them from a workflow step with uses, passing inputs via with and reading outputs from ${{ .steps.<id>.outputs.<name> }}. See Actions for the full pattern.

Notebooks​

Scheduled notebook

Run a Jupyter notebook on a cron schedule. Demonstrates how to pass a file between two jobs via H2O Drive (the runner filesystem is not shared between jobs).

Create AI Engines​

Launch Driverless AI

Create a DAI engine, wait for STATE_RUNNING, and return its API URL as an action output.

Launch H2O-3

Create an H2O-3 cluster via AI Engine Manager, wait until it is ready, and return its API URL as an action output.

LLM calls​

LLM call (h2oGPTe)

Run a single h2oGPTe LLM call as an inline step — direct inference, structured JSON output, or agent mode.

Notifications​

Post to Slack

Send a Slack message via an incoming webhook (passed in as a secret).

Send email

Send an email through any SMTP server.

  • Actions — defining, using, and publishing composite actions.
  • Reusable workflows — the workflow: job type, for logic that needs its own jobs and sandbox.
  • Runner environment — what's available inside the runner (platform token, pre-installed tools, how to install Python clients).
  • Secrets — how to supply secrets such as Slack webhooks and SMTP passwords.
Feedback