Reading Tracks
Five ways through these docs — understand it, click it, call it, extend it, or run it. Every page declares which one it belongs to.
These docs are organised by subject — connectors, strategies, agents, billing. Most readers arrive with an audience instead: they want to understand the model, or drive the product, or extend the engine, or keep it running. Almost every subject section serves two or three of those at once.
So rather than split the tree, every page declares the one audience that cannot do its job without it. Pick the track that matches why you're here.
Understand it
The model — what a thing is and why it works that way. Start here if the vocabulary is new.
Click it
Build routes in the visual editor, without writing code or calling an endpoint.
Call it
Drive Ductor over REST, Connect-RPC, or the CLI. The largest track.
Extend it
Write Go, TypeScript, or Python against the engine — SDKs, strategies, providers.
Run it
Deploy, migrate, monitor, secure, and recover a running instance.
Tracks are a lens, not a location
A page's track is metadata, not its address. connectors/connections stays
where it is; it simply announces that it is written for someone calling the
API. Nothing moved, and no link you have already saved has changed.
That also means the tracks overlap by design. A page can name a secondary
audience, and many do — an explanatory page that ends with a worked curl
call belongs to Understand it, and points at Call it as well.