June 2026

The Add tool palette on an agent node now lists the Alarm, Filesystem, and Tool nodes alongside integrations and MCP server. Pick one and it is created, wired into the agent, and opened to its own config step, so you no longer have to drop the node on the canvas and draw the edge by hand.

Hover any automation node to find two new buttons on its left edge. Pin saves the node's last output as mock data, so steps that follow keep running without re-executing it, and Disable skips the node at run time. Both stay highlighted while active, so you can see and undo them without hovering.

The HTTP Request node now handles almost any API call without leaving NoClick. Add headers and query parameters as key/value rows, choose a body type (None, JSON, Form URL-Encoded, or Raw) on any method so even GET and DELETE can send a body, and set a timeout, retries with backoff, and redirect and SSL options. Authentication adds an API key in a header or the query string, a custom header, and OAuth2 client credentials alongside Bearer and Basic auth.

In a hurry, click Import cURL at the top of the node and paste a curl command from any API's docs, or paste the command straight onto the canvas to get a pre-configured node. By default a 4xx or 5xx response fails the node loudly so the run stops with a clear error. Turn on Continue on HTTP Error to branch on the error instead.

Building a workflow in chat now shows what is happening as it happens. A live counter shows tokens as they stream, and a node ledger lists each node as it is added and configured, with the field it is setting. Click a node in the ledger to jump to it on the canvas. If a build fails, its message stays put with a Retry button instead of vanishing.

The Runs pill in the top left of the canvas is now always there, even before this session's first run. It shows the latest run's status and when it happened, and clicking it opens the results popup for that run. Inside the popup, a run switcher lets you load any earlier run and see what each node produced, scrolling back through your run history without leaving the canvas.

The results popup now also opens on its own after more kinds of runs. A Run from here run that starts partway through the flow, and runs started by a trigger like a schedule or a webhook, both open the popup the same way a full run does.

If a workflow only starts on its own, like on an incoming email, a new row, a schedule, or a webhook, pressing Run now explains that instead of doing a surprise one-off run. The popup lists what each trigger watches and the address or URL that fires it. From there you can Add a Run step to add a manual entry point, Open config for any trigger, or Run anyway for a single manual run.

Finding the right action is now fuzzy and ranked. Search by any word order, by an abbreviation, or by words from the operation's description, and the closest match jumps to the top, highlighted so you can press Enter to pick it. The same ranked search now powers the agent tool allowlist and every searchable dropdown.

Multi-value list fields, like a message's To, Cc, and Bcc or a list of feed URLs, now save as you type. A fresh empty box appears below the moment you fill the last one, so you no longer press Enter to add each value.

Lists no longer loop silently: one click converts a list reference into a visible iteration node on the canvas.