What are the main parts of Ansible's automation engine, and what flows into and out of it?

The engine has an inventory, modules, plugins and an API. Users and playbooks feed it, as do cloud and CMDB sources. It acts on hosts and network devices.

* Clouds and CMDBs supply host information, users run playbooks, and the engine puts hosts and network devices into the described state. *

Inputs:

• Users start runs and write playbooks, which describe what to do.
• Public/private clouds and a CMDB supply host information, typically through dynamic inventory.

Inside the engine:

• Inventory: which hosts exist and how they're grouped.
• Modules: the tools tasks use to change systems.
• Plugins: extensions for connections, inventory sources, output and more.
• API: lets other systems drive Ansible programmatically.

Outputs: the engine connects to hosts (servers) and networking devices and puts them into the described state.

Go deeper:

• How Ansible Works (Ansible Collaborative / Red Hat) — inventory, modules, plugins and the Python API, plus cloud-sourced dynamic inventory.
• Ansible (software) (Wikipedia) — architecture overview.