Independent Argo Workflows resource

Argo Workflows | Kubernetes Workflow Automation

Argo Workflows is a Kubernetes-native workflow engine designed to automate containerized tasks, coordinate complex processes, and manage parallel job execution. It enables developers and DevOps teams to build scalable workflows using YAML-based configurations, DAG structures, and reusable templates. From CI/CD automation to data processing and machine learning pipelines, Argo Workflows simplifies cloud-native workflow orchestration.

10free working tools
12in-depth guides
100%runs in your browser
Argo Workflows 4.1 – an octopus conductor orchestrating a DAG of parallel workflow tasks
The basics

What is Argo Workflows?#

Argo Workflows is an open-source, container-native workflow engine that turns Kubernetes into a platform for orchestrating parallel jobs. A workflow is a Kubernetes custom resource; each step runs as a pod; and a controller inside your cluster schedules steps, passes data between them, retries failures and records results.

Because it is built from Kubernetes primitives, Argo Workflows inherits namespaces, RBAC, resource quotas, autoscaling and GitOps tooling — a natural choice for data pipelines, machine learning, CI jobs and infrastructure automation.

  • Every step is an isolated container with its own image and resources
  • Steps and DAG tasks run in parallel whenever dependencies allow
  • Declarative YAML that works with kubectl, Helm and Argo CD
Read the full introduction
argo CLI / kubectlsubmit · get · logs Argo Server + UIport 2746 · SSO Kubernetes APIWorkflow CRDs Workflowcontroller pod: extract pod: train pod: report Artifact repository (S3 · GCS · Azure · MinIO) creates pods
Features

Argo Workflows features for production pipelines#

The main capabilities of the engine, grouped by what they control: structure, data, scheduling, reliability, security and visibility.

01

Templates

Reusable units of work: container, script, resource, HTTP, suspend, steps and DAG templates.

02

DAGs & steps

Express ordering as sequential stages or as a dependency graph with rich depends logic.

03

Retries & deadlines

Backoff, retry policies and activeDeadlineSeconds keep pipelines resilient.

04

Artifacts

Pass files between steps through S3, GCS, Azure Blob, MinIO and more.

05

CronWorkflows

Schedule pipelines with multiple cron expressions, timezones and concurrency policies.

06

Parameters & loops

Fan out with withItems and withParam; branch with when.

07

Security

Per-template service accounts, RBAC, SSO and non-root execution with emissary.

08

Observability

Built-in Prometheus metrics, a live UI, log archiving and a SQL workflow archive.

Argo Workflows tools

Try the Argo Workflows tools right here#

Three tools work directly on this page: generate a Workflow manifest, validate YAML against Argo Workflows rules, and explain a cron schedule. Everything runs in your browser. Use the button below to open any of the ten full tools.

The manifest includes a service account, a one-hour deadline, a one-day TTL, pod garbage collection and resource requests.

Download

Download the Argo Workflows v4.1.6 SBOM#

A software bill of materials (SBOM) lists every package inside a release. Security teams use it to check images for known vulnerabilities and licence issues before running them.

argo-workflows-v4.1.6-sbom.tar.gzSPDX 2.2 · tar.gz · 278 KB · 6 documents

This archive is the SBOM published with the official Argo Workflows v4.1.6 release. It contains one SPDX document per component image and one for the web UI's JavaScript dependencies.

SHA-2560b18e32aeecd82bc61d1272fe9724143535441c3c45d5b1aebe1036d26f3ee2d

How teams use this file

  • Scan the listed packages for known vulnerabilities (CVEs) before upgrading.
  • Check open-source licences for compliance and audit reports.
  • Keep a record of exactly which component versions run in each cluster.

What is inside

SPDX documentPackages
argocli.spdx
argo CLI and Argo Server image
37
workflow-controller.spdx
Workflow controller image
35
argoexec.spdx
Executor image
46
argoexec-nonroot.spdx
Executor image, non-root variant
37
argo-workflows-crdinstaller.spdx
CRD installer image
33
bom-yarn.spdx
Web UI JavaScript dependencies
3,919

Verify and scan

sha256sum argo-workflows-v4.1.6-sbom.tar.gz
tar -xzf argo-workflows-v4.1.6-sbom.tar.gz
grype sbom:./dist/argocli.spdx
Getting started

Run your first Argo Workflows pipeline in four steps#

A first installation has four stages: a Kubernetes cluster, the workflow controller, the argo command-line tool and a submitted workflow.

  1. 1

    Prepare a cluster

    Any conformant cluster works. For learning, create one locally with kind, k3d or minikube.

    $ kind create cluster
  2. 2

    Install the controller

    Apply the release manifest to an argo namespace, or use the Helm chart for GitOps.

    $ kubectl apply -n argo -f install.yaml
  3. 3

    Install the CLI

    Download the argo binary to submit, watch and inspect workflows.

    $ argo version
  4. 4

    Submit a workflow

    Generate YAML with our tools and watch it run.

    $ argo submit hello.yaml --watch
Argo Workflows YAML

Copy-ready YAML examples#

Four common workflow shapes: sequential steps, a DAG, a scheduled CronWorkflow and a reusable WorkflowTemplate. Each manifest runs on a standard installation.

argo-workflows-examples.yaml
apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: etl-steps-
spec:
  entrypoint: main
  templates:
    - name: main
      steps:
        - - name: extract
            template: say
            arguments: {parameters: [{name: msg, value: extract}]}
        - - name: transform-a        # these two run in parallel
            template: say
            arguments: {parameters: [{name: msg, value: transform-a}]}
          - name: transform-b
            template: say
            arguments: {parameters: [{name: msg, value: transform-b}]}
        - - name: load
            template: say
            arguments: {parameters: [{name: msg, value: load}]}
    - name: say
      inputs: {parameters: [{name: msg}]}
      container:
        image: busybox:1.36
        command: [echo, "{{inputs.parameters.msg}}"]
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
  name: argo-executor
rules:
  - apiGroups: [argoproj.io]
    resources: [workflowtaskresults]
    verbs: [create, patch]

Since v3.4, workflow pods need this permission to report outputs. Bind it to the service account named in spec.serviceAccountName.

Kubernetes integration

Configure Kubernetes the right way#

Most production issues are really Kubernetes issues: a missing role, a pod that cannot be scheduled, a secret in the wrong place. Get the foundations right and Argo Workflows becomes a dependable platform.

  • A dedicated service account with the minimal executor role
  • Resource requests, memory limits and namespace quotas
  • Secrets from Kubernetes Secrets or workload identity — never parameters
  • Restricted pod security and non-root containers
Use cases

Argo Workflows use cases#

Because every step runs in its own container with its own resources, the same engine handles machine learning, batch data, CI and infrastructure jobs. Each card shows a typical pipeline and the features it relies on.

ML

Machine learning pipelines

Training pipelines combine steps with very different needs. Data preparation runs on CPU nodes, training requests GPUs through resource limits and node selectors, and evaluation compares metrics before a model is registered. Each step is a separate pod, so it gets exactly the hardware it needs.

Read the related guide →

Typical pipeline

  1. Prepare data
  2. Train (GPU)
  3. Evaluate
  4. Register model

Argo Workflows features used

  • withItems for hyperparameter sweeps
  • Artifacts for datasets and models
  • when gate on accuracy
ETL

Data processing & ETL

Batch jobs usually split input into partitions. A script step lists the partitions as JSON, withParam creates one pod per partition, and a final step merges the results. The cluster autoscaler adds nodes for the burst and removes them when the workflow finishes.

Read the related guide →

Typical pipeline

  1. List partitions
  2. Process ×N
  3. Merge
  4. Publish

Argo Workflows features used

  • withParam fan-out
  • parallelism limits
  • retryStrategy with backoff
CI

CI and build automation

Build pipelines benefit from parallelism: linting, unit tests and image builds can run at the same time once the code is checked out. Argo Events can submit a workflow from a Git webhook, and an exit handler reports the result back.

Read the related guide →

Typical pipeline

  1. Checkout
  2. Lint + Test
  3. Build image
  4. Scan

Argo Workflows features used

  • DAG depends logic
  • Shared volume or artifacts
  • onExit notifications
OPS

Infrastructure automation

Resource templates create, patch or delete Kubernetes objects and wait for conditions such as a Job succeeding. Combined with CronWorkflows, this covers database backups, certificate checks, environment cleanup and scheduled reports.

Read the related guide →

Typical pipeline

  1. Schedule
  2. Create Job
  3. Wait for status
  4. Clean up

Argo Workflows features used

  • CronWorkflow schedules
  • resource templates
  • successCondition checks
Argo CLI

The Argo Workflows command chain#

The argo CLI follows the life of a workflow: submit it, list runs, inspect the node tree, read logs and retry failed nodes.

FAQ

Argo Workflows frequently asked questions#

Short answers about the engine, how it compares to other tools, and this website.

Ask another question
What is Argo Workflows?

Argo Workflows is an open-source, container-native workflow engine for Kubernetes. Each step runs in its own pod, and workflows are defined as Kubernetes custom resources in YAML.

Is this Argo Workflows website the official Argo Project site?

No. This Argo Workflows website is an independent educational resource. It is not affiliated with, endorsed by or sponsored by the Argo Project, the CNCF or the Linux Foundation. The official documentation lives at argo-workflows.readthedocs.io.

Are the tools free to use?

Yes. Every tool on this site is free, requires no sign-up and runs entirely in your browser, so the YAML you paste is never uploaded to a server.

Which Argo Workflows version do the guides cover?

Guides target the current 4.x release line and point out differences from 3.x where they matter, such as the CronWorkflow schedules list and plural mutexes and semaphores.

Do I need Kubernetes experience to start?

Basic familiarity with pods, namespaces and kubectl helps. Our guides, starting with What Is Argo Workflows?, explain the Kubernetes concepts you need along the way.

What is the difference between steps and a DAG?

Steps run in sequential stages with parallel items inside each stage. A DAG declares dependencies per task, so each task starts as soon as its own dependencies finish.

Can I schedule workflows?

Yes. CronWorkflows run workflows on cron schedules with timezone, concurrency and history settings. Our CronWorkflow generator builds them for you.

How do I report an error in a guide?

Use the contact page. We review every correction request and update the article, noting significant changes in its update date.