Skip to main content
Runs an expensive subgraph once and reuses its result on later runs. Wire the work you want to cache — typically “load a big list and build a lookup map” — to the Miss output port. That subgraph must end with an End node and write its result into the configured output assign. The first time a given cache key is seen the Miss subgraph runs and its output is cached; every later run with the same key skips straight out Hit with the cached value already in the output assign.

​
Configuration

  • Cache key — a literal key with {assign} interpolation, e.g. manager-{manager_name}. This is usually what you want. Leave it blank to key off a single assign’s raw value instead (below).
  • Cache key assign — used only when Cache key is blank: the assign whose value keys the cache (handy when the key is a whole struct).
  • Output assign — the Miss subgraph writes here; on a hit the cached value is placed here for you.
  • TTL (seconds) — optional. How long a cached entry lives before it is recomputed. Leave empty to use the cache’s default cleanup.

​
Caching scope & staleness

The cache is keyed by (org, flow, key value) — so two cache nodes in the same flow with the same key share a value, and other flows are isolated. It’s shared across all runs on this server for the TTL window, with no manual invalidation. If the underlying data can change mid-batch, use a short TTL — or fold a freshness marker (e.g. today’s date) into the cache key — so stale lookups expire quickly.

​
Ports