Data Actions

Everything Flowfile can do to your data is one of the actions on this page — read a file, filter rows, join two tables, summarize by group, train a model, write the result somewhere. You drag one onto the canvas, fill in its settings, and connect it to the next one. This page lists all of them, so you can see the full set without opening the app and find the one you need without knowing which category it lives in.

On the canvas, an action is called a node

The palette, the canvas and the rest of this reference use the word node. One node is one action, and the two words mean the same thing here.

Find an action by what you want to do

You want to	Use	Where
Keep only the rows that meet a condition	Filter data	Transformations
Add a calculated column	Formula	Transformations
Remove duplicate rows	Drop duplicates	Transformations
Fix blanks, stray spaces and inconsistent casing	Data cleansing	Transformations
Rename a lot of columns at once	Rename columns	Transformations
Choose, reorder or drop columns	Select data	Transformations
Number the rows	Add record Id	Transformations
Split one cell's list into separate rows	Text to rows	Transformations
Look up values from another table (VLOOKUP)	Join	Combine Operations
Stack two files that share the same columns	Union data	Combine Operations
Match names that are spelled slightly differently	Fuzzy match	Combine Operations
Run part of the flow only when a condition holds	Gate	Combine Operations
Reuse another flow as a single step	Run Flow	Combine Operations
Total or average per customer, month, region	Group by	Aggregations
Turn row values into columns (crosstab)	Pivot data	Aggregations
Turn columns back into rows	Unpivot data	Aggregations
Running total, rank or moving average	Window functions	Aggregations
Count the rows	Count records	Aggregations
Read every file in a folder	List files	Input Sources
Look at the data before deciding what to do	Explore data	Output Operations
Save a result colleagues can query and chart	Write to Catalog	Output Operations
Predict a number or a category	Train Model, then Apply Model	Machine Learning
Write SQL or Python instead of filling in a form	SQL Query, Polars code, Python Script	Transformations

Nothing here fits? Build your own node in the Node Designer, or install one someone else published.

The six categories

The palette groups actions the same way this reference does, under the same headings.

Input Sources

10 actions. Get data in: files, folders, databases, cloud storage, REST APIs, Kafka, Google Analytics, the catalog.
Transformations

14 actions. Reshape one dataset: filter, sort, cleanse, calculate columns, or drop into SQL, Polars or Python.
Combine Operations

8 actions. Bring datasets together with a join, union or fuzzy match — group connected records, branch the flow with a gate, or call another flow.
Aggregations

5 actions. Summarize and restructure: group, pivot, unpivot, count, window calculations.
Machine Learning

4 actions. Split a dataset, fit a model, score new rows, and measure how well it did.
Output Operations

7 actions. Send results out: files, databases, cloud storage, the catalog, an API response — or explore them on screen.

Every action, A to Z

48 actions as of 2026-09. The palette is the live list; this table is generated from the same source (flowfile_core/flowfile_core/configs/node_store/nodes.py) and each name matches what the palette shows.

Action	What it does	Category	Lite
Add record Id	Generate unique identifiers for each row	Transformations	●
API response	Return this dataset as the body of an HTTP API endpoint	Output
Apply Model	Score data with a trained model	Machine Learning
Count records	Calculate the total number of rows	Aggregations	●
Cross join	Create all possible combinations between two datasets	Combine	●
Data cleansing	Fix nulls, whitespace, unwanted characters and casing in one step	Transformations
Drop duplicates	Remove duplicate rows based on selected columns	Transformations	●
Evaluate Model	Compare actual vs predicted columns and compute quality metrics	Machine Learning
Explore data	Interactive data exploration and analysis	Output	●
Filter data	Keep only rows that match your conditions	Transformations	●
Flow Input	Named entry point for data when this flow runs inside another flow	Input
Flow Output	Named exit point exposing this dataset when the flow runs inside another flow	Output
Formula	Create or modify columns using custom expressions	Transformations	●
Fuzzy match	Join datasets based on similar values instead of exact matches	Combine
Gate	Pass data through only when a condition holds; otherwise skip what follows	Combine
Google Analytics	Load reports from a Google Analytics 4 property	Input
Graph solver	Group related records in graph-structured data	Combine
Group by	Aggregate data by grouping and calculating statistics	Aggregations	●
Join	Merge two datasets based on matching column values	Combine	●
Kafka Source	Read data from a Kafka or Redpanda topic	Input
List files	List a folder's contents as a table	Input
Manual input	Create data directly	Input	●
Multi-field formula	Apply one expression to many columns at once	Transformations
Pivot data	Convert data from long format to wide format	Aggregations	●
Polars code	Write custom Polars DataFrame transformations	Transformations	●
Python Script	Execute Python code on an isolated kernel container	Transformations
Random Split	Randomly partition rows into named groups (e.g. train/test)	Machine Learning
Read data	Load data from CSV, Excel, Parquet and other files	Input	●
Read from Catalog	Read a table from the data catalog	Input	●
Read from cloud provider	Read data from AWS S3 and other cloud storage	Input
Read from Database	Load data from database tables or queries	Input
Rename columns	Bulk-rename columns by prefix, suffix, or a formula	Transformations	●
REST API	Read JSON data from a REST API with auth and pagination	Input
Run Flow	Execute a flow from the catalog, mapping data and parameters into it	Combine
Select data	Choose, rename, and reorder columns to keep	Transformations	●
Sort data	Order your data by one or more columns	Transformations	●
SQL Query	Write SQL queries against connected data sources	Transformations
Take Sample	Work with a subset of your data	Transformations	●
Text to rows	Split text into multiple rows based on a delimiter	Transformations
Train Model	Fit a regression or classification model	Machine Learning
Union data	Stack multiple datasets by combining rows	Combine	●
Unpivot data	Transform data from wide format to long format	Aggregations	●
Wait For	Pass the left input through; the right input only enforces ordering	Combine
Window functions	Rolling, cumulative, rank, tile and partition-aggregate calculations	Aggregations
Write data	Save your data as CSV, Excel, Parquet and other files	Output	●
Write to Catalog	Save data as a table in the data catalog	Output	●
Write to cloud provider	Save data to AWS S3 and other cloud storage	Output
Write to Database	Save data to database tables	Output

A ● marks the 22 actions that also run in Flowfile Lite, the browser-only edition. Lite adds two of its own — External Data and External Output, which fetch from and post to a URL — for 24 in total.

How an action works

Every node on the canvas behaves the same way:

Inputs and outputs. A node reads from whatever is connected to its input handles and passes its result on from its output handle. The count is fixed per action: Join takes two inputs, Random Split emits two outputs, Write data has no output at all.
Settings. Click a node and its settings open in the right-hand panel. The fields differ per action; each section below lists them.
Schema preview. Once configured, a node reports its output columns and types without running the flow. Run it to see actual rows in the preview panel.
Lazy by default. Most actions build up a Polars query that only executes when you run the flow, so intermediate steps cost nothing until you ask for a result.

New to the canvas? Building Flows covers creating, connecting, configuring and running nodes. Every action here is also available from Python.