The Graph

Every automation is a directed graph stored in your library. Understanding the two wire types is the most important concept — everything else follows from it.

Flow wires vs data wires

Flow wires answer "what runs next?" Data wires answer "what value does this input need?"

Four node kinds

Triggers

Exactly one trigger per automation. Triggers have a single flow output and expose typed data outputs from the event (message, member, channel, and so on).

See Triggers for the full list.

Actions

Actions perform work — send a template, create a ticket, add a role, and more. They have one flow input and one flow output, plus data inputs for the values they need.

Actions only run when reached by a flow wire. See Action nodes.

Control-flow nodes

Meta nodes branch, filter, delay, or stop execution: If, Filter, Delay, Switch, For Each, and Stop.

See Control flow.

Value nodes

Values, fetchers, number/text transforms, dictionaries, and template builders produce typed values. They have no flow pins — they are evaluated only when a downstream node needs their output, then memoized for the rest of the run.

You do not wire flow into value nodes. Wire their outputs into action or control-flow data inputs instead.

Execution model

1. The trigger fires and flow walks downward through connected actions.
2. When an action needs a value, Assistance walks backward along data wires and evaluates any value nodes required.
3. Filter silently ends the run when its condition is false.
4. Stop always ends the run immediately.
5. For Each repeats its Each branch once per array item. Item and Index outputs are only valid inside that branch.
6. Continuation Action buttons start a new run on an On Press branch — they do not continue the current flow.

Graph limits

Runtime step budgets and nesting limits depend on your plan. See Publishing.

Related