n8n

🧠 Free Recall ❓ Quiz 🔍 Cues 🔮 Simulation 🥷 ShadowBox 📐 Encode Cards

A workflow is a collection of nodes connected together to automate a process.

A workflow is a collection of nodes connected together to automate a process. You build workflows on the workflow canvas.

You may need to run your workflow manually when building and testing, or if your workflow doesn't have a trigger node.

All new workflows are inactive by default.
You need to activate workflows that start with a trigger node or Webhook node so that they can run automatically. When a workflow is inactive, you must run it manually.
To activate or deactivate your workflow, open your workflow and toggle Inactive / Active.
Once a workflow is active, it runs whenever its trigger conditions are met.

Nodes are the key building blocks of a workflow. They perform a range of actions, including:

Starting the workflow.
Fetching and sending data.
Processing and manipulating data.

n8n provides a collection of built-in nodes, as well as the ability to create your own nodes. Refer to:

Built-in integrations to browse the node library.
Community nodes for guidance on finding and installing community-created nodes.
Creating nodes to start building your own nodes.

When you add a node to a workflow, n8n displays a list of available operations. An operation is something a node does, such as getting or sending data.
There are two types of operation:

Triggers start a workflow in response to specific events or conditions in your services. When you select a Trigger, n8n adds a trigger node to your workflow, with the Trigger operation you chose pre-selected. When you search for a node in n8n, Trigger operations have a bolt icon .
Actions are operations that represent specific tasks within a workflow, which you can use to manipulate data, perform operations on external systems, and trigger events in other systems as part of your workflows. When you select an Action, n8n adds a node to your workflow, with the Action operation you chose pre-selected.

Node settings#
The node settings under the Settings tab allow you to control node behaviors and add node notes.
When active or set, they do the following:

Request Options: Select Add Option to view and select these options.Batching: Control how to batch large numbers of input items.
Ignore SSL Issues: Download the response even if SSL validation isn't possible.
Proxy: Use this if you need to specify an HTTP proxy.
Timeout: Set a timeout for the request in ms.
Always Output Data: The node returns an empty item even if the node returns no data during execution. Be careful setting this on IF nodes, as it could cause an infinite loop.
Execute Once: The node executes once, with data from the first item it receives. It doesn't process any extra items.
Retry On Fail: When an execution fails, the node reruns until it succeeds.
On Error:Stop Workflow: Halts the entire workflow when an error occurs, preventing further node execution.
Continue: Proceeds to the next node despite the error, using the last valid data.
Continue (using error output): Continues workflow execution, passing error information to the next node for potential handling.

Connections

A connection establishes a link between nodes to route data through the workflow. A connection between two nodes passes data from one node's output to another node's input.

Executions

An execution is a single run of a workflow.

Execution modes#
There are two execution modes:

Manual: run workflows manually when testing. Select Test Workflow to start a manual execution. You can do manual executions of active workflows, but n8n recommends keeping your workflow set to Inactive while developing and testing.
Production: a production workflow is one that runs automatically. To enable this, set the workflow to Active.

Manual executions#
Manual executions allow you to run workflows directly from the canvas to test your workflow logic. These executions are "ad-hoc": they run only when you manually select the Execute workflow button.
Manual executions make building workflows easier by allowing you to iteratively test as you go, following the flow logic and seeing data transformations. You can test conditional branching, data formatting changes, and loop behavior by providing different input items and modifying node options.

When performing manual executions, you can use data pinning to "pin" or "freeze" the output data of a node. You can optionally edit the pinned data as well.
On future runs, instead of executing the pinned node, n8n will substitute the pinned data and continue following the flow logic. This allows you to iterate without operating on variable data or repeating queries to external services. Production executions ignore all pinned data.

Partial executions#
Clicking the Execute workflow button at the bottom of the workflow in the Editor tab manually runs the entire workflow. You can also perform partial executions to run specific steps in your workflow. Partial executions are manual executions that only run a subset of your workflow nodes.
To perform a partial execution, select a node, open its detail view, and select Execute step. This executes the specific node and any preceding nodes required to fill in its input data. You can also temporarily disable specific nodes in the workflow chain to avoid interacting with those services while building.
In particular, partial executions are useful when updating the logic of a specific node since they allow you to re-execute the node with the same input data.

Troubleshooting partial executions#
Some common issues you might come across when running partial executions include the following:

The destination node is not connected to any trigger. Partial executions need a trigger.
This error message appears when you try to perform a partial execution without connecting the workflow to a trigger. Manual executions, including partial executions, attempt to mimic production executions when possible. Part of this includes requiring a trigger node to describe when the workflow logic should execute.
To work around this, connect a trigger node to the workflow with the node you're trying to execute. Most often, a manual trigger is the simplest option.

Please execute the whole workflow, rather than just the node. (Existing execution data is too large.)
This error can appear when performing partial executions on workflows with large numbers of branches. Partial executions involve sending data and workflow logic to the n8n backend in a way that isn't required for full executions. This error occurs when your workflow exceeds the maximum size allowed for these messages.
To work around this, consider using the limit node to limit node output while running partial executions. Once the workflow is running as intended, you can disable or delete the limit node before enabling production execution.

Production executions#
Production executions occur when a triggering event or schedule automatically runs a workflow.
To configure production executions, you must attach a trigger node (any trigger other than the manual trigger works) and switch workflow's toggle to Active. Once activated, the workflow automatically executes whenever the trigger condition occurs.

A dirty node is a node that executed successfully in the past, but whose output n8n now considers stale or unreliable. They're labeled like this to indicate that if the node executes again, the output may be different. It may also be the point where a partial execution starts from.

Why n8n marks nodes dirty#
There are several reasons why n8n might flag execution data as stale. For example:

Inserting or deleting a node: labels the first node that follows the inserted node dirty.
Modifying node parameters: labels the modified node dirty.
Adding a connector: labels the destination node of the new connector dirty.
Deactivating a node: labels the first node that follows the deactivated node dirty.

n8n saves workflows in JSON format. You can export your workflows as JSON files or import JSON files into your n8n library. You can export and import workflows in several ways.

You can copy and paste a workflow or parts of it by selecting the nodes you want to copy to the clipboard (Ctrl + c or cmd +c) and pasting it (Ctrl + v or cmd + v) into the Editor UI.
To select all nodes or a group of nodes, click and drag:

Credentials are securely stored authentication information used to connect n8n workflows to external services such as APIs, or databases.

Key concepts
Flow logic

Splitting workflows with conditional nodes#
Splitting uses the IF or Switch nodes. It turns a single-branch workflow into a multi-branch workflow. This is a key piece of representing complex logic in n8n.

Merging data#
Merging brings multiple data streams together. You can achieve this using different nodes depending on your workflow requirements.

Merge data from different data streams or nodes: Use the Merge node to combine data from various sources into one.
Merge data from multiple node executions: Use the Code node for complex scenarios where you need to merge data from multiple executions of a node or multiple nodes.
Compare and merge data: Use the Compare Datasets node to compare, merge, and output data streams based on the comparison.

Merge data from different data streams#
If your workflow splits, you combine the separate streams back into one stream.
Here's an example workflow showing different types of merging: appending data sets, keeping only new items, and keeping only existing items. The Merge node documentation contains details on each of the merge operations.

Merge data from different nodes#
You can use the Merge node to combine data from two previous nodes, even if the workflow hasn't split into separate data streams. This can be useful if you want to generate a single dataset from the data generated by multiple nodes.

The Compare Datasets node compares data streams before merging them. It outputs up to four different data streams.

Looping is useful when you want to process multiple items or perform an action repeatedly, such as sending a message to every contact in your address book. n8n handles this repetitive processing automatically, meaning you don't need to specifically build loops into your workflows. There are some nodes where this isn't true.

n8n nodes take any number of items as input, process these items, and output the results. You can think of each item as a single data point, or a single row in the output table of a node.

Creating loops#
n8n typically handles the iteration for all incoming items. However, there are certain scenarios where you will have to create a loop to iterate through all items. Refer to Node exceptions for a list of nodes that don't automatically iterate over all incoming items.

Loop until a condition is met#
To create a loop in an n8n workflow, connect the output of one node to the input of a previous node. Add an IF node to check when to stop the loop.
Here is an example workflow that implements a loop with an IF node:

Loop until all items are processed#
Use the Loop Over Items node when you want to loop until all items are processed. To process each item individually, set Batch Size to 1.
You can batch the data in groups and process these batches. This approach is useful for avoiding API rate limits when processing large incoming data or when you want to process a specific group of returned items.
The Loop Over Items node stops executing after all the incoming items get divided into batches and passed on to the next node in the workflow so it's not necessary to add an IF node to stop the loop.

Waiting#
Waiting allows you to pause a workflow mid-execution, then resume where the workflow left off, with the same data. This is useful if you need to rate limit your calls to a service, or wait for an external event to complete. You can wait for a specified duration, or until a webhook fires.
Making a workflow wait uses the Wait node. Refer to the node documentation for usage details.
n8n provides a workflow template with a basic example of Rate limiting and waiting for external events.

Sub-workflows#
You can call one workflow from another workflow. This allows you to build modular, microservice-like workflows. It can also help if your workflow grows large enough to encounter memory issues. Creating sub-workflows uses the Execute Workflow and Execute Sub-workflow Trigger nodes.
Sub-wokflow executions don't count towards your plan's monthly execution or active workflow limits.

Error handling#
When designing your flow logic, it's a good practice to consider potential errors, and set up methods to handle them gracefully. With an error workflow, you can control how n8n responds to a workflow execution failure.

In n8n, all data passed between nodes is an array of objects. It has the following structure:

Code#
You can use the Code node to manipulate binary data in your workflows. For example, Get the binary data buffer: get the binary data available in your workflow.
Binary data#
Binary data is any file-type data, such as image files or documents.
This page collects resources relating to binary data in n8n.

Working with binary data in your workflows#
You can process binary data in n8n workflows. n8n provides nodes to help you work with binary data. You can also use code.

Nodes#
There are three key nodes dedicated to handling binary data files:

Read/Write Files from Disk to read and write files from/to the machine where n8n is running.
Convert to File to take input data and output it as a file.
Extract From File to get data from a binary format and convert it to JSON.
There are separate nodes for working with XML and HTML data:

HTML
XML
And nodes for performing common tasks:

Compression
Edit Image
FTP
You can trigger a workflow based on changes to a local file using the Local File trigger.
To split or concatenate binary data items, use the data transformation nodes.

Item linking concepts#
Each output item created by a node includes metadata that links them to the input item (or items) that the node used to generate them. This creates a chain of items that you can work back along to access previous items. This can be complicated to understand, especially if the node splits or merges data. You need to understand item linking when building your own programmatic nodes, or in some scenarios using the Code node.
Data item linking#
An item is a single piece of data. Nodes receive one or more items, operate on them, and output new items. Each item links back to previous items.
You need to understand this behavior if you're:

Building a programmatic-style node that implements complex behaviors with its input and output data.
Using the Code node or expressions editor to access data from earlier items in the workflow.
Using the Code node for complex behaviors with input and output data.

Data mapping#
Data mapping means referencing data from previous nodes.
Function#
A function is a block of code designed to perform a certain task. In n8n, you can write custom JavaScript or Python code snippets to add, remove, and update the data you receive from a node.
The Code node gives you access to the incoming data and you can manipulate it. With this node you can create any function you want using JavaScript code.
To transform this kind of structure into the n8n data structure you can use the data transformation nodes:

Aggregate: take separate items, or portions of them, and group them together into individual items.
Limit: remove items beyond a defined maximum number.
Remove Duplicates: identify and delete items that are identical across all fields or a subset of fields.
Sort: organize lists of in a desired ordering, or generate a random selection.
Split Out: separate a single data item containing a list into multiple items.
Summarize: aggregate items together, in a manner similar to Excel pivot tables.

Transforming data#
n8n uses a predefined data structure that allows all nodes to process incoming data correctly.
Your incoming data may have a different data structure, in which case you will need to transform it to allow each item to be processed individually.
For example, the image below shows the output of an HTTP Request node that returns data incompatible with n8n's data structure. The node returns the data and displays that only one item was returned.
Data item processing#
Nodes can process multiple items.
For example, if you set the Trello node to Create-Card, and create an expression that sets Name using a property called name-input-value from the incoming data, the node creates a card for each item, always choosing the name-input-value of the current item.
For example, this input will create two cards. One named test1 the other one named test2:

Unstarred

Dunk (exclude from study)

n8n

Review Questions

Todos

Sensemaker