Chapter 1
What is an agent?
Define AI agents, distinguish them from chatbots and workflows, and learn the vocabulary used by the rest of the playbook.
Interactive HTML playbook
A progressive, Python-focused guide that starts with agent concepts and builds toward Microsoft Agent Framework components, orchestration, observability, and hosting.
Use the chapter grid to move from foundational agent vocabulary into hands-on MAF component shells. Chapter authors can add content inside each prepared section without changing the site chrome.
🤖 Curious how this playbook was built? See the interactive agent-orchestration wireframe →
Chapter 1
Define AI agents, distinguish them from chatbots and workflows, and learn the vocabulary used by the rest of the playbook.
Chapter 2
Connect common agentic patterns to MAF's component families before introducing APIs.
Chapter 3
Create a Python MAF agent with a clear role, instructions, and model client boundary.
Chapter 4
Understand how MAF separates agent behavior from provider chat clients, messages, and conversation state.
Chapter 5
Add safe, well-scoped tools so an agent can act beyond text generation.
Chapter 6
Model repeatable multi-step agent systems with workflow executors, edges, and sequential or concurrent paths.
Chapter 7
Choose handoff and group collaboration patterns when one agent is not enough.
Chapter 8
Make longer-running agent workflows observable, resumable, and controllable by people.
Chapter 9
Add cross-cutting behavior and OpenTelemetry traces without burying operational concerns inside agent logic.
Chapter 10
Understand how MAF projects move from local authored components toward YAML configuration, hosted execution, and interactive debugging.