Sandbox
@githubnext/gh-aw-workshop

Workshop for GitHub Agentic Workflows in Actions

This is a guided workshop for building a GitHub Agentic Workflows setup from scratch. It shows how to define workflows in Markdown, connect GitHub context and triggers, run them with `gh-aw`, and iterate on the result. The workshop focuses on a useful end state: a scheduled repository status report that runs in GitHub Actions, plus patterns you can reuse for documentation upkeep, PR review, and similar tasks.

47 stars18 forksJavaScriptUpdated 7d ago
Who it's for

Builders who want to learn GitHub Agentic Workflows and use them in GitHub Actions.

What it delivers

You can build and adapt agentic GitHub automations without starting from toy examples.

What it does

End-to-end workshop path

Lesson files in `workshop/` take you from setup to workflow building, running, debugging, and extension.

Workflow pattern examples

Later lessons cover PR review, conditional logic, data sources, MCP tools, memory, sub-agents, evaluation, and orchestration.

Supporting GitHub config

`.github/aw`, `.github/agents`, `.github/skills`, and `.github/mcp.json` provide the workshop’s agentic workflow scaffolding.

Doc build and validation scripts

`scripts/build-docs.js` and related tests generate and check the workshop site and assets.

README

GitHub Agentic Workflows Factory Tour

[!IMPORTANT] GitHub Agentic Workflows and the gh-aw CLI are in public preview. You may notice product changes, rough edges, or minor differences from the screenshots and examples in this workshop.

Mona's Agent Factory

What you will learn to build

Build a real AI-powered GitHub workflow from scratch and leave with a safe production-style agentic automation you can adapt to your own repositories.

Sample daily repository status report generated by the finished workflowSample daily repository status report generated by the finished workflow

Why this workshop

Most workflow tutorials stop at toy examples. This workshop takes you all the way to a useful end result: an agentic workflow that runs in GitHub Actions, inspects a repository, and publishes a daily status report automatically.

By the time you finish, you will know how to:

  • define an agentic workflow in Markdown
  • use the gh-aw CLI to validate, run, and iterate on workflows
  • connect triggers, prompts, and GitHub context into one practical automation
  • adapt the pattern for scheduled reporting, documentation upkeep, PR review, and more

What you'll build

You'll build a scheduled, AI-powered repository status report that runs in GitHub Actions and posts a useful summary without custom shell-script orchestration.

Preview of a completed workflow run in the GitHub Actions UIPreview of a completed workflow run in the GitHub Actions UI

Who this is for

You are...You'll get...
A developer new to agentic workflowsA clear, end-to-end path from zero to a working automation
A GitHub Actions userA practical mental model for when agentic workflows are a better fit than classic YAML and scripts
A platform or DevOps engineerA reusable pattern for AI-assisted repository triage with auditable runs in GitHub
An evaluator, mentor, or workshop hostA hands-on artifact that makes agentic workflow concepts easy to demonstrate and assess

What makes this different from a classic Actions tutorial

Instead of spending most of your time wiring together scripts, conditionals, and API calls, you define the workflow in a Markdown brief and let the agent reason over the repository at runtime.

That means you focus on:

  • the task you want completed
  • the repository context the agent should use
  • the trigger that should run the workflow
  • the output that should be published back to GitHub

Ready to begin?

:arrow_right: Start with workshop/00-welcome.md


:copyright: GitHub • Code of ConductMIT License

[!NOTE] Microsoft and GitHub facilitators: Request a demo environment at aka.ms/golden-ticket.

Files in the repo

Repository payload13 top-level entries
  • .devcontainer
  • .github
  • .markdownlint
  • .vscode
  • scripts
  • workshop
  • .gitattributes
  • .gitignore
  • .markdownlint-cli2.yaml
  • AGENTS.md
  • package-lock.json
  • playwright.svg-contrast.config.js
  • README.md

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More tutorials & guides

shareAI-lab/
learn-claude-code

Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1

77k
luongnv89/
claude-howto
luongnv89/claude-howtoTutorials & Guides

A visual, example-driven guide to Claude Code — from basic concepts to advanced agents, with copy-paste templates that bring immediate value.

41k
agentskills/
agentskills
agentskills/agentskillsTutorials & Guides

Specification and documentation for Agent Skills

25k