Chaos Toolkit

Chaos Toolkit is an open-source chaos engineering toolkit for developers who want to express system experiments as code. Its strongest safeguard is a steady-state check that can stop an experiment before its method runs; teams seeking a managed platform should look elsewhere.

Overview

The toolkit lets teams define experiments in JSON or YAML, keep them alongside other code, and use them for collaboration and orchestration. Each experiment states a steady-state hypothesis and probes to test it, then specifies a method of actions and probes. If the system fails the steady-state check, the method does not run and the experiment must bail out. That makes the design a sensible fit for teams that want a clear precondition before injecting faults, though the safety gate does not replace careful experiment design.

Key features

Experiments and extensibility

Experiments can be embedded in a CI/CD chain, making the toolkit relevant to teams that want chaos checks to become part of their delivery workflow. Its Open API supports extensions for different systems, and users can add extensions after installing the CLI and core library. The extension directory includes integrations for AWS, Azure, Google Cloud Platform, Kubernetes, Kafka, Slack, Datadog, Grafana, OpenTelemetry and Prometheus.

Kubernetes support, cloud and network fault injection, experiment scheduling and blast-radius controls are supported. Together, these capabilities cover several common needs for controlled experiments, while the code-oriented model favors developers comfortable maintaining experiment definitions and a Python-based CLI.

Runtime and deployment

The standalone CLI is implemented in Python 3, officially supports Python 3.8 and later, and has been tested only against CPython. The installation guide covers Python setup on macOS, Debian or Ubuntu, and Windows, with the CLI installed using pip. Teams using other Python implementations should account for that testing scope.

Chaos Toolkit is self-hosted and open source under Apache License 2.0. Reliably is a separate platform, with SaaS, private SaaS and on-premises options; those deployment choices are not part of the standalone CLI.

Pricing

Chaos Toolkit: free. The plan is the open-source CLI, free forever under Apache License 2.0. It suits teams able to install and run a self-hosted tool and manage their own experiments and extensions. There is no paid tier or managed service within this plan; teams wanting a platform offering can consider Reliably separately.

Platforms

The toolkit lists API, Linux, macOS, self-hosted, web and Windows platforms. Its installation guide specifically covers macOS, Debian or Ubuntu, and Windows, and describes a Python-and-pip CLI installation. The listed platforms make it broadly relevant across environments, but the supported runtime remains Python 3.8+ with CPython as the only tested implementation.

Who it's for

Chaos Toolkit is best for development and operations teams that want free, self-hosted chaos experiments stored and orchestrated like code, particularly when a steady-state gate and CI/CD embedding suit their workflow. It is less suitable for buyers who want a hosted service bundled with the toolkit, or teams unwilling to maintain a Python CLI and its extensions.

Pros and cons

• Pros: JSON or YAML experiment definitions can live alongside code, supporting collaboration and orchestration in familiar workflows.
• Pros: A failed steady-state check prevents the experiment method from running, providing a meaningful safety gate.
• Pros: The CLI is free forever and Apache-licensed, with extensions spanning major cloud, infrastructure, messaging and observability tools.
• Cons: The standalone offering is self-hosted, so teams seeking SaaS or managed operations need another product.
• Cons: The CLI requires Python 3.8 or later, and only CPython has been tested, narrowing confidence for other Python runtimes.
• Cons: Users install extensions after the CLI and core library, adding setup and maintenance work for the facets they need.

Alternatives

For a broader comparison, see Chaos Engineering Platforms.

• ChaosBlade is another free option; consider it if its support for three system platforms, four programming-language applications and 200+ experiment scenarios better matches your target systems.
• Reliably is the alternative to consider if you want a freemium platform and its open-source, no-lock-in approach.
• Chaos Mesh is a free, open-source cloud-native platform that runs on Kubernetes; pick it when a Kubernetes-specific platform is the priority.
• PowerfulSeal is a free, self-hosted Kubernetes chaos-testing option installed with pip or Docker; consider it for that narrower setup.
• Toxiproxy is another free option, with Windows, macOS and Linux platforms.
• chaosd is a free option for Linux.
• FaultMesh offers a freemium web option.
• Krkn is a free option for Linux.

Verdict

Choose Chaos Toolkit if your team wants a free, open-source way to define self-hosted chaos experiments as code and values a steady-state check that can halt an experiment before its method runs. Its main trade-off is operational ownership: installation, runtime compatibility and extensions are yours to manage, and the CLI is not a hosted platform.