A living, self-maintaining book about agentic workflows, agent orchestration, and agentic scaffolding.
This book doesn't just teach about agentic workflows—it uses them! The book automatically:
- 📝 Processes community suggestions from GitHub issues
- ✍️ Updates content using AI agents
- 📚 Maintains markdown sources and generates a PDF build
- 🌐 Publishes to GitHub Pages
- 📰 Creates blog posts for larger curated updates
- Online: https://arivero.github.io/agentbook (GitHub Pages)
- PDF: Download from GitHub Actions artifacts or the repo PDF
- Source: Browse the book directory
The book covers practical patterns for agentic workflows, safe orchestration, skills and tooling, and real-world automation in GitHub and beyond. Browse the source in the book/ directory or read it online for the full flow.
- Community Input: Open an issue with a suggestion
- Automated Intake: A standard workflow posts an acknowledgment and applies
acknowledged - Agentic Routing: A GH-AW routing workflow classifies the suggestion
- Agent Processing:
triaged-fast-trackfor small low-risk fixestriaged-for-researchfor larger/ambiguous updates (then research -> phase 1 -> phase 2 -> assignment)
- Automatic Build: PDF build and site deployment workflows run
- Publishing: Changes deploy to GitHub Pages
We welcome contributions! Here's how:
The primary way to contribute is by opening an issue with your suggestion:
- Open an issue using our Content Suggestion template
- Automated Processing: Our GitHub Agentic Workflows (GH-AW) will analyze your suggestion
- Follow Updates: The issue moves through labels:
acknowledged→triaged-fast-trackortriaged-for-research→assigned(orrejected). Slow-track phases are tracked via agent comments.
What happens to suggestions:
- ✅ Accepted (fast-track): Routed with
triaged-fast-track, implemented by Copilot, PR opened, and issue closed - ✅ Accepted (research lane): Routed with
triaged-for-research, researched, receives phase 1 and phase 2 outputs, markedassigned, then closed - ❌ Rejected/Out of Scope: Any agent can reject with rationale and close at any stage
All suggestions related to agentic workflows, orchestration, scaffolding, skills/tools, or GitHub agents are welcome.
To run the book locally:
# Clone the repository
git clone https://github.com/arivero/agentbook.git
cd agentbook
# Serve with Jekyll (requires Ruby)
bundle install
bundle exec jekyll serve
# Or use GitHub Pages locally
gem install github-pages
github-pages serveagentbook/
├── book/ # Book content
│ ├── README.md # Book introduction
│ └── chapters/ # Manuscript sources
├── blog/ # Blog posts
├── _layouts/ # Jekyll layouts
├── _posts/ # Blog post files
├── .github/
│ ├── workflows/ # GitHub Actions
│ └── ISSUE_TEMPLATE/ # Issue templates
├── _config.yml # Jekyll configuration
└── index.md # Homepage
This repository uses GitHub Actions and GH-AW workflows for publishing, validation, and issue-processing.
For workflow lifecycle details, see WORKFLOW_PLAYBOOK.md. For required operator setup (tokens, permissions, and operating checklist), use content.md. This is the guide for the required human operator.
This work is licensed under the MIT License and available for educational purposes.
- README - Project overview and workflow summary
- content - Required human-operator guide
- SETUP - Installation and configuration guide
- WORKFLOW_PLAYBOOK - Fast-track and full-playbook maintenance process
- LICENSE - MIT License
This project demonstrates the power of agentic workflows by being a living example of the concepts it teaches.
Note: This is an experimental project exploring self-maintaining documentation through agentic workflows.