Argo Workflows Cheat Sheet

Updated 2026-05-22

Next Topic: Backstage Developer Portal Cheat Sheet

🧠Study flashcards on this topic135 cards · spaced repetition→

Argo Workflows is a Kubernetes-native workflow engine that defines workflows as Directed Acyclic Graphs (DAGs) or sequential steps using Custom Resource Definitions (CRDs). Each workflow step runs in its own container, making it ideal for CI/CD pipelines, ML training, data processing, and any containerized batch workload. This cheat sheet covers the full workflow specification, template types, artifact management, concurrency, observability, and the broader Argo ecosystem.

What This Cheat Sheet Covers

This topic spans 17 focused tables and 149 indexed concepts, 135 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Core Concepts and ArchitectureTable 2: Template TypesTable 3: Workflow Specification EssentialsTable 4: Parameters and VariablesTable 5: ArtifactsTable 6: Conditional Execution and Control FlowTable 7: Retry StrategiesTable 8: Synchronization and ConcurrencyTable 9: Lifecycle ManagementTable 10: Artifact Repository ConfigurationTable 11: CronWorkflowsTable 12: Argo Events IntegrationTable 13: Common PatternsTable 14: Observability and MetricsTable 15: Security and Access ControlTable 16: Hera Python SDK and Client LibrariesTable 17: Argo Project Ecosystem and Comparisons

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Core Concepts and Architecture

Argo Workflows runs inside Kubernetes and manages workflows as CRDs. Understanding the key components — controller, server, executor, and CRD kinds — is the foundation for all subsequent concepts.

Concept	Example / Value	Description
Workflow CRD	`kind: Workflow`	The core Kubernetes custom resource representing a single workflow execution instance with a unique name.
WorkflowTemplate CRD	`kind: WorkflowTemplate`	A reusable, namespace-scoped template library that other workflows can reference via `templateRef`.
ClusterWorkflowTemplate CRD	`kind: ClusterWorkflowTemplate`	Like WorkflowTemplate but cluster-scoped, accessible from any namespace.
CronWorkflow CRD	`kind: CronWorkflow`	• Triggers a Workflow on a cron schedule • supports multiple schedules, timezone, and concurrencyPolicy
Workflow Controller	`workflow-controller` Deployment	• Reconciliation loop watching Workflow CRDs • schedules pods and reconciles one workflow at a time per controller
Argo Server	`argo-server` on port 2746	HTTP(S) API server that serves the Argo UI, REST API, and artifact proxy.
argoexec	volume-mounted binary	• Binary injected into every step pod • acts as the main entrypoint, wraps the user container, and handles artifact I/O
Emissary Executor	default since v3.4	• The only supported executor • does not require privileged access • works on GKE Autopilot • exits with code 64 on executor failure
Pod Anatomy	init, main, wait	Every step pod has three containers: `init` (fetch params/artifacts), `main` (user container), `wait` (save outputs and clean up).
Argo Server Auth Modes	`client`, `server`, `hybrid`	Controls how the server authenticates API calls: using the caller's token, a static server token, or falling back to server token if no client token is provided.

Table 2: Template Types

Argo Workflows supports nine template types. Each type serves a different purpose — from running containers and scripts to orchestrating other templates with DAGs or steps.

Template Type	Example / Value	Description
container template	`template: {container: {image: python:3.11}}`	• Runs a single container • the most common template type • equivalent to a Kubernetes Pod spec

The rest of this section is part of the full cheat sheet

Table 3: Workflow Specification Essentials

The Workflow manifest uses spec fields to define entry points, arguments, service accounts, metadata, and execution policies. These fields control how a workflow starts and where it runs.

Field / Pattern	Example / Value	Description
entrypoint	`spec.entrypoint: main`	Names the template that is executed first when the workflow starts.

The rest of this section is part of the full cheat sheet

Table 4: Parameters and Variables

Argo Workflows supports a rich variable system for passing data between templates. Variables use {{simple}} or {{= expression}} syntax and are resolved at submission time or runtime.

Variable / Pattern	Example / Value	Description
inputs.parameters	`inputs.parameters: [{name: message}]`	• Declares input parameters for a template • referenced inside the template as `{{inputs.parameters.message}}`.

The rest of this section is part of the full cheat sheet

Table 1: Core Concepts and Architecture

Concept	Example / Value	Description
Workflow CRD	`kind: Workflow`	The core Kubernetes custom resource representing a single workflow execution instance with a unique name.
WorkflowTemplate CRD	`kind: WorkflowTemplate`	A reusable, namespace-scoped template library that other workflows can reference via `templateRef`.
ClusterWorkflowTemplate CRD	`kind: ClusterWorkflowTemplate`	Like WorkflowTemplate but cluster-scoped, accessible from any namespace.
CronWorkflow CRD	`kind: CronWorkflow`	• Triggers a Workflow on a cron schedule • supports multiple schedules, timezone, and concurrencyPolicy
Workflow Controller	`workflow-controller` Deployment	• Reconciliation loop watching Workflow CRDs • schedules pods and reconciles one workflow at a time per controller
Argo Server	`argo-server` on port 2746	HTTP(S) API server that serves the Argo UI, REST API, and artifact proxy.
argoexec	volume-mounted binary	• Binary injected into every step pod • acts as the main entrypoint, wraps the user container, and handles artifact I/O
Emissary Executor	default since v3.4	• The only supported executor • does not require privileged access • works on GKE Autopilot • exits with code 64 on executor failure
Pod Anatomy	init, main, wait	Every step pod has three containers: `init` (fetch params/artifacts), `main` (user container), `wait` (save outputs and clean up).
Argo Server Auth Modes	`client`, `server`, `hybrid`	Controls how the server authenticates API calls: using the caller's token, a static server token, or falling back to server token if no client token is provided.

Table 2: Template Types

Argo Workflows supports nine template types. Each type serves a different purpose — from running containers and scripts to orchestrating other templates with DAGs or steps.

Template Type	Example / Value	Description
container template	`template: {container: {image: python:3.11}}`	• Runs a single container • the most common template type • equivalent to a Kubernetes Pod spec

Table 3: Workflow Specification Essentials

The Workflow manifest uses spec fields to define entry points, arguments, service accounts, metadata, and execution policies. These fields control how a workflow starts and where it runs.

Field / Pattern	Example / Value	Description
entrypoint	`spec.entrypoint: main`	Names the template that is executed first when the workflow starts.

Table 4: Parameters and Variables

Argo Workflows supports a rich variable system for passing data between templates. Variables use {{simple}} or {{= expression}} syntax and are resolved at submission time or runtime.

Variable / Pattern	Example / Value	Description
inputs.parameters	`inputs.parameters: [{name: message}]`	• Declares input parameters for a template • referenced inside the template as `{{inputs.parameters.message}}`.