> For the complete documentation index, see [llms.txt](https://developer.harness.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://developer.harness.io/resilience-testing/chaos-engineering/new-to-chaos-engineering/overview.md).

# Overview

Chaos Engineering is the practice of proactively introducing controlled faults into applications and infrastructure to test the resilience of business services. Developers, QA teams, performance engineers, and Site Reliability Engineers (SREs) run chaos experiments to measure system resilience and discover weaknesses before they impact production.

Harness Chaos Engineering provides end-to-end tooling for resilience testing at enterprise scale through proven chaos engineering principles.

{% embed url="<https://www.youtube.com/watch?v=dk8WPek1P-w>" %}

### What you will learn from this topic

This topic introduces the capabilities, use cases, and deployment modes of Harness Chaos Engineering.

### Core capabilities

Harness Chaos Engineering provides the following capabilities:

* **Chaos experiments**: Run chaos experiments with more than 200 built-in faults, probes, and actions. These cover Kubernetes, cloud platforms, Linux, Windows, and application runtimes.
* **Resilience probes**: Use probes to programmatically observe the expected behavior or steady state. Integrate with application performance monitoring (APM) tools and applications.
* **Actions**: Perform custom tasks within a chaos experiment. Use actions for notifications, webhooks, and load-testing scripts.
* **Enterprise governance**: Use ChaosGuard to control who runs experiments, where, and when.
* **Centralized chaos execution plane**: Use scalable architecture with centralized execution and distributed agents through Harness Delegate.
* **Connectors**: Integrate with CI/CD pipelines, monitoring tools, and cloud service providers.
* **AI-powered capabilities**: Use the AI Reliability Agent for experiment-creation, optimization, and failure-resolution recommendations.
* **MCP tools**: Use Harness MCP server tools from AI editors, such as Claude Desktop, Windsurf, and Cursor.
* **GameDay portal**: Run controlled production experiments to validate incident-response procedures and system recovery.

The platform includes RBAC, single sign-on (SSO), comprehensive logging, and audit capabilities. It is available in SaaS and on-premises deployments. Go to [Harness Platform key concepts](/harness-ai/new-to-harness-platform/overview.md) to understand general Harness Platform concepts and features.

### Use cases

Harness Chaos Engineering supports the following use cases:

* **Resilience testing in deployment pipelines**: Add chaos experiments to deployment pipelines for continuous resilience validation.
* **Load testing with resilience testing**: Run chaos experiments with load-testing tools under traffic stress.
* **GameDay exercises**: Run controlled production tests to validate incident-response procedures and recovery.
* **Disaster recovery testing**: Validate backups, failover mechanisms, and recovery procedures through fault injection.

### Deployment modes

Choose one of the following deployment modes:

* [SaaS](/resilience-testing/chaos-engineering/new-to-chaos-engineering/on-premise-vs-saas.md#saas): Use a managed cloud service with automatic updates and scaling.
* [On-premises](/resilience-testing/chaos-engineering/new-to-chaos-engineering/on-premise-vs-saas.md#on-premise): Deploy in your infrastructure for complete control.

### Chaos fault library

Browse more than 200 ready-to-use chaos faults across your infrastructure.

Go to [Chaos Faults](/resilience-testing/chaos-engineering/faults/chaos-fault-categories/chaos-faults-reference.md) to browse the fault library.

### New Chaos Studio

{% hint style="info" %}
New Chaos Studio features

Harness Chaos Engineering offers the **New Chaos Studio** experience. Your studio version depends on your onboarding date:

The following options are available:

* **New Chaos Studio**: Available if you onboarded on or after August 21, 2025.
* **Old Chaos Studio**: Available if you onboarded before August 21, 2025.

The New Chaos Studio includes the following capabilities:

* New Chaos Studio: Design chaos experiments with a streamlined workflow.
* Timeline view: View experiment execution and results on a timeline.
* [Experiment-level probes](/resilience-testing/chaos-engineering/use-chaos-engineering/probes/experiment-level-probes.md): Configure probes at the experiment level.
* [Actions](/resilience-testing/chaos-engineering/use-chaos-engineering/actions.md): Run custom operations, delays, and scripts during experiments.
* [ChaosHubs across scopes](/resilience-testing/chaos-engineering/use-chaos-engineering/chaoshub.md): Manage ChaosHubs with flexible scoping.
* Runtime variable support: Use dynamic variables during experiment execution.
* [Templates](/resilience-testing/chaos-engineering/use-chaos-engineering/templates.md): Reuse fault, probe, and action templates.
* [Custom faults](/resilience-testing/chaos-engineering/faults/custom-faults.md): Create and manage custom fault definitions.

Contact your Harness support representative to access New Chaos Studio features.
{% endhint %}