Resources · n8n integration

n8n Call n8n Workflow Tool nodeConfigure Call n8n Workflow Tool in n8n.

An agent can call one of your own workflows instead of an external API. The n8n Call n8n Workflow Tool node packages any set of nodes as a tool, and exposes 3 fields: Name, Description and Source. For builders who already have workflows worth reusing.

Try the Call n8n Workflow Tool node on n8nSee the 3 parameters

Verified Trustpilot reviews · AI, automation & growth agency

Why automate

What does the n8n Call n8n Workflow Tool node do?

It turns another n8n workflow into a tool an agent can call. A tool, in an AI context, is an add-on resource the model reaches for when it needs information or an action it cannot produce on its own. The node runs that second workflow, waits, and hands its output data back to the agent as the tool result. Anything you can build with n8n nodes becomes callable that way.

Start with the case that comes up most: an agent that answers questions about data living somewhere n8n already reads. You build a small workflow that queries the source and returns rows, publish it, then point this node at it with Source set to Database. The agent asks, the sub-workflow runs, the rows come back. That pattern is exactly what the template AI chat with any data source builds, workflow 2026 on the n8n site, with Hacker News and a Code node behind the tool.

Second case: one agent, several jobs. Rather than wiring a dozen action nodes onto the agent itself, each job gets its own workflow and its own tool entry. The Description is what routes the request, since it tells the model when to reach for this tool and nothing else does. Template 3025 runs on that idea, with a chatbot that keeps long-term memory and routes to different sub-workflows. Building a tool around Gmail or Slack costs you one workflow each.

Third case: reuse. A workflow that formats a report from Google Sheets already exists and runs on a schedule. The same workflow becomes a tool without being copied, so one fix lands in both places. Template 3770 pushes this further and packages a set of workflows behind a single agent.

Prefer a dedicated node when the agent only needs one call against a service that already has one. An OpenAI or Perplexity tool node is one node, not two workflows. This node earns its place when the work takes several steps, branches, or logic you would rather not rebuild inside an agent.

Known limits, from the node itself. It has no sub-node port of its own: it attaches to an AI Agent and is driven by it. With Database as the source and the agent running outside the manual chat test panel, the sub-workflow has to be published, or the call fails with Workflow is not active and cannot be executed. That error text is returned to the agent as the tool result, so it hides inside a long answer. The page covers version 2, the latest in the catalogue; an older workflow may show fewer options.

Parameters

Which fields matter in the Call n8n Workflow Tool node?

The Call n8n Workflow Tool node has 3 parameters. For each one: the node as you configure it in n8n, what the parameter changes, and our field notes.

01

Name

name

What you see in n8n

Notes & use cases

This is the function name the agent sees in its list of available tools, so it is what the model calls when it decides to use you. Letters, numbers and underscores only, no spaces.

Key parameters

  • Name: the name of the function to be called, letters, numbers and underscores only; the field shows e.g. My_Color_Tool as its placeholder.
Use cases
name the tool after the job, not the workflow file, so a model reading a list of five entries picks the right one on the first try.
02

Description

description

What you see in n8n

Notes & use cases

Write the instruction that tells the agent when to use this tool, and what shape the input should take. The model reads nothing else about your workflow, so a vague line here means a tool that never fires or fires at the wrong moment.

Key parameters

  • Description: custom text aimed at the agent; the placeholder shows the register expected, Call this tool to get a random color. The input should be a string with comma separated names of colors to exclude.
Use cases
spell out the trigger condition and the input format in two sentences, the way the placeholder does.
03

Source

source

What you see in n8n

Notes & use cases

Where n8n gets the workflow it runs. Pick the stored workflow, or paste one in.

Key parameters

  • Source: database (Database) loads the workflow from the database by ID, parameter (Define Below) passes the JSON code of a workflow.
  • Workflow: required with Database, the selector that points at the sub-workflow.
  • Workflow Inputs: required, maps what the tool call sends into the sub-workflow fields.
  • Workflow JSON: required with Define Below, the workflow JSON code to execute.
Use cases
keep Database for anything you maintain, and remember the sub-workflow must be published once the agent leaves the test panel.
Need help

Need help automating Call n8n Workflow Tool with n8n?

A person reads every message.

FAQ

Call n8n Workflow Tool and n8n, the usual questions

01Is the n8n Call n8n Workflow Tool node free to use?
Yes. It ships inside the n8n AI package, @n8n/n8n-nodes-langchain, which comes with n8n itself. Nothing to install, nothing extra to pay on the n8n side, and it behaves the same on n8n Cloud, the hosted offer run by n8n, and on a self-hosted instance under the Community Edition and its Sustainable Use licence. A workflow built in one place runs in the other. The one cost that exists sits elsewhere: the provider behind the chat model your agent uses, OpenAI, Anthropic, Google or Mistral, bills its own API calls under its own terms. n8n adds nothing to that bill, and this node calls no provider by itself.
02What credentials does the Call n8n Workflow Tool node need?
None. There is no credential and no Authentication selector on this node, which is why the page has no connection steps. It calls a workflow that already lives in your n8n instance, so there is no account to link and no key to paste. The credential you do need sits one node away: the chat model plugged into the agent carries the provider key, and the sub-workflow you call carries whatever credentials its own nodes use. Setting this up is a matter of filling three fields and picking a workflow, not of opening a provider dashboard.
03What are the limits of the Call n8n Workflow Tool node?
Three worth knowing. It has no sub-node port of its own, so nothing plugs into it; it runs as an action node on each incoming item and is driven by the agent it is attached to. With Database as the source and the agent running in production rather than the manual chat test panel, the sub-workflow has to be published, otherwise the call fails with Workflow is not active and cannot be executed. That error text goes back to the agent as the tool result, so it can disappear inside a long answer: if the agent says it could not retrieve data, check the sub-workflow is published and that its execution list shows a run. Last, this page describes version 2, the highest in the catalogue.
04What do you need to plug in to make it work?
An agent and a chat model, at minimum. This node is a tool, meaning an add-on resource the model reaches for during a request, so it does nothing alone: it attaches to an AI Agent node, which is the root node receiving the workflow items, while sub-nodes like the chat model, the memory or a tool hang off its ports, one type of sub-node per port. Start small. Build the sub-workflow first, publish it, set Source to Database, pick it in the Workflow selector, map Workflow Inputs, then write the Description so the model knows when to call it. Test from the chat panel before running it for real.
05n8n or Make for calling a workflow as a tool?
It depends on where you want the automation to live. Make, formerly Integromat, is a hosted automation platform with no self-hosting option and a price model counted per operation, which reads well when volumes are low and predictable. n8n gives you both paths, the hosted n8n Cloud or your own server, so the data an agent passes through a sub-workflow stays where you decide. The visual logic differs too, and the honest way to choose is to look at hosting, control over the data, the cost model and which canvas you read faster. Both run the scenarios described here.
Hack'celeration Lab

Get our weekly integration tips.

No spam. Unsubscribe anytime.

Same platform

Other docs you might need on this stack.

Browse all integrations