Skip to content

Preset Catalog

The platform ships a catalog of presets — ready-made trigger kinds for sources and execution profiles for sinks. You pick a preset, configure it, and place it; you never create your own types.

A source is a named trigger endpoint: a preset plus its connection config and a placement.

PresetWhat it doesPlacement
WebhookAn HTTP POST to the source’s address creates an event. The address is shown in the flow editor once the flow is savedPlatform or a source node
Folder watchA file appearing or changing in watched paths (local folder or SMB share) creates an event. Paths are configured per flow on the WHEN line. The platform’s own processing artifacts (.lock and .track files written by file actions) never fire the triggerPlatform or a source node
MQTTA message on a subscribed topic creates an event. The broker address and credentials live on the source; topics are per flowPlatform or a source node
Every day at…A cron expression fires the flow on a schedule. The cron lives on the source — editing the source updates every flow using itPlatform or a source node
Azure EventGridMessages from an Azure Event Grid topic subscription become eventsPlatform
MCPAI agents using the Model Context Protocol create events directly — see the MCP Server guideA source node

Sources are inherently location-bound: a folder watcher watches that folder, a webhook listens on that address. That is why a source is pinned to one source node or runs on the platform — there is no “any node” for sources.

The connection config on the New source slide-over renders per param type: secret params (webhook signing secret, SMB/MQTT credentials) are a dropdown of your tenant’s secrets — the stored value is the secret’s URI, resolved when the config ships to a node — and true/false params (the webhook ALLOW_* security switches) are switches, not text fields. Everything else is free text.

A sink is a named execution target. It exists only when it pins or constrains:

  • Pinned to a node — steps on this sink always run on that node.
  • Any matching node — the step runs on whichever of your sink nodes covers its constraints. Add attribute constraints (for example zone=ot) to narrow the pool.

Bare “any matching node” and “platform” are always available choices in the flow editor’s placement picker — they are never stored as sinks.

Nodes carry an attribute sheet: reported capabilities (os, arch, engine version — written by the node at registration, read-only) and user-assigned attributes (for example zone, tools). Define your own attributes with their value lists on the Nodes screen; values are sanitized and deduplicated, and renaming a value updates every node and constraint at once.

For every step, the hub checks one thing: the step’s constraints ⊆ a node’s attributes.

  • The step’s needs fall out of the action’s library requirements (declared by the platform library).
  • The sink’s constraints add attribute filters.
  • “Any matching” picks one node from the matching pool and fails over inside the pool; a pinned node never falls back to the platform, and work with no match is held and flagged on the flow.