Run a chain and read the results

Hit Run. Enlace works out which nodes depend on which — through connections and through field mappings — and fires them in the right order, straight from your browser to your target API.

Independent steps really do run in parallel

If two nodes don't depend on each other or on anything the other depends on, Enlace fires both requests at the same time instead of waiting for one to finish first. In the demo chain, that's what makes updating a customer and creating a product happen together instead of one after another — the two requests genuinely overlap in flight, not just run in a permissive order.

A node still waits for everything it depends on — directly or through a mapped field — no matter how many other things are running at the same time.

Reading the Results pane

Every node in the chain gets a row in the Results pane (bottom of the canvas) before the run even starts, in dependency order, with a status icon that updates live as the run proceeds — pending, in flight, completed, failed, or skipped. Expand a row to see the exact request sent and the response received.

A presets collection (Wait/Assert) gets one row like any other node — expand it to see each preset inside it settle in order, underneath the collection's own row.

Bearer/Basic credentials are redacted here (see Authenticate your requests for exactly what is and isn't), so most runs are safe to leave open while pairing or screen-sharing — but check that page if you're using an API key sent as a header.

The pane stays empty until you actually run something — nodes sitting on an untouched canvas don't get placeholder rows. Clear empties the list again once you're done with it, but keeps the underlying run result around in the background so any field mappings pulling a preview from it keep resolving; it's the visible rows that get wiped, not the data those mappings need.

The canvas locks while a run is in progress

Field values, credentials, and connections can't be edited while a run is going — the inspector and canvas controls disable themselves, and even a direct store edit is rejected, so nothing can silently "take" without actually affecting the run in progress. You can still drag nodes around to rearrange the canvas; position was never part of what gets executed.

If a step fails

Whatever hasn't started yet is halted — no more requests fire after the point of failure. Anything already in flight in parallel with the failed step still finishes and reports its own result, since a request that's already gone out to your API can't be called back. Nothing downstream of the failure runs.

Rerunning after a failure: "Rerun failed"

Once a run has left anything unfinished — a failure, or nodes that never got reached — a small Rerun failed icon appears right next to Run (hover it for the tooltip; it's icon-only, not a labeled button). It skips every node that completed cleanly last time and picks back up from wherever the chain actually stopped, instead of firing the whole thing again from the top.

Debug gets the same treatment: a Debug failed icon appears next to it whenever Rerun failed does. It's the same resume behavior, but honors breakpoints for whatever's left — so you can jump straight to pausing right before the node that actually failed, instead of re-running the whole chain in Debug mode to get there. See Debug a chain step by step.

This matters most partway through a long chain: fix whatever made step 6 fail, hit Rerun failed, and steps 1–5 aren't re-sent — their prior responses are reused as-is, including for anything downstream mapping a field from them.

Edit a node that already completed, though, and that safety net doesn't apply to it: Enlace compares each completed node's config against what it was at the start of the last run, and a change — to that node, or to anything upstream of it — makes it (and everything downstream of it) rerun fresh rather than reuse a result that config change might have invalidated. Only genuinely untouched nodes get skipped.

The last run's result survives a page refresh too (this browser's own autosave), so Rerun failed is still there if you reload — it's only cleared once a different workflow loads.

A cycle in your chain

If a set of nodes end up depending on each other in a loop — A needs B, B needs A — Enlace refuses to run any of it and tells you before sending a single request, rather than executing part of the chain and hanging.

Want to pause mid-run and inspect a step before it fires?

See Debug a chain step by step — arm a breakpoint on a connector and use Debug instead of Run.