Overview

Experiments are stored as JSON or YAML, which lets teams collaborate on them and orchestrate them like other code. Each experiment sets a steady-state hypothesis, uses probes to check it, then defines a method of actions and probes. If the system does not meet the steady-state condition, the method is withheld and the experiment must stop. That is a useful guardrail before injecting faults, though it does not replace careful planning of what an experiment may affect.

The project is licensed under Apache 2.0 and offered as a standalone open-source CLI. That keeps the toolkit free to run under a self-hosted deployment model, but teams must manage the runtime and environment themselves. Reliably is a separate offering, with SaaS, private SaaS and on-premises options.

Key features

Experiments, automation and integrations

Chaos Toolkit can be embedded in a CI/CD chain, so teams can make experiments part of a delivery workflow. Its Open API allows extension for different systems; after installing the CLI and core library, users can add extensions for particular facets of chaos engineering. Integrations include AWS, Azure, Google Cloud Platform, Kubernetes, Kafka, Slack, Datadog, Grafana, OpenTelemetry and Prometheus.

The toolkit supports Kubernetes, cloud fault injection, network fault injection, experiment scheduling and blast-radius controls. That combination gives teams a broad set of ways to exercise system behavior, while the extension model means setup depends on installing the relevant components rather than expecting every integration in the core CLI.

Python CLI and setup

The CLI is implemented in Python 3, officially supports Python 3.8 and later, and has only been tested against CPython. Installation uses pip, with setup guidance for macOS, Debian or Ubuntu, and Windows. This is a reasonable route for teams already equipped to manage Python tooling; it is a meaningful constraint for organizations that want a packaged, centrally operated service instead.

The project invites issue reports, issue voting, stars, community Slack participation, pull requests and extension proposals. Its contribution guide also recommends enabling two-factor authentication on contributor accounts to reduce risk if an account is compromised.

Pricing

Chaos Toolkit — free. The plan is the open-source CLI, free forever under the Apache License 2.0. There is no paid tier in this standalone toolkit plan; users install the CLI and core library, then add extensions as needed. Its cost advantage comes with self-hosted operation and Python-based setup rather than managed service convenience.

Platforms

Chaos Toolkit is available for API, Linux, macOS, self-hosted, web and Windows environments. Its installation guidance covers macOS, Debian or Ubuntu, and Windows, while the CLI's Python 3.8+ requirement and CPython-only testing should guide compatibility planning.

Who it's for

Choose Chaos Toolkit if your team wants versionable experiment definitions, CI/CD integration and a free, extensible toolkit it can operate itself. It particularly suits developers and reliability teams comfortable with Python and with selecting and installing extensions for their target systems. Look elsewhere if you need a managed deployment, or if maintaining a Python CLI is not a fit for your team.

Pros and cons

Pros: JSON or YAML experiments can live and evolve alongside application code, supporting collaboration and orchestration.
Pros: The steady-state check blocks the experiment method when its precondition fails, providing a concrete safety gate.
Pros: The free-forever Apache-licensed CLI supports CI/CD use and can be extended for a range of cloud, infrastructure, observability and messaging systems.
Cons: The offering is a self-hosted CLI, not a managed service; teams must own its installation and operating environment.
Cons: Python 3.8+ and CPython-only testing narrow the supported runtime expectations compared with tools that do not require this CLI setup.
Cons: Extensions are installed separately after the CLI and core library, adding a setup step for teams with specific integration needs.

Alternatives

For a broader comparison, browse Chaos Engineering Platforms.

Consider FaultMesh if you want a freemium option with a free Starter plan for up to 3 services and a paid Growth plan at 49.00 USD per month, billed per h.

Reliably is a separate option if you want its stated open-source approach and a choice of SaaS, private SaaS or on-premises deployment.

ChaosBlade is another free open-source chaos engineering tool, described as supporting 3 system platforms, 4 programming-language applications and 200+ experiment scenarios.

Krkn is another free chaos engineering option.

Choose Chaos Mesh if you want a free, open-source chaos engineering platform that runs on Kubernetes.

LitmusChaos offers both a free hosted SaaS plan and a free open-source platform for installation on your own cluster.

Toxiproxy is a free, open-source TCP proxy under the MIT license.

Pumba is a free, open-source container chaos testing tool under Apache License 2.0.

Verdict

Chaos Toolkit is a compelling choice for developers and reliability teams who want free, code-oriented experiments with a clear steady-state gate and the flexibility to extend their setup. Its self-hosted Python CLI is also its main trade-off: teams that want managed operations or less runtime ownership should choose another platform.

	Chaos Toolkit	Krkn-AI
Free plan	Yes	Yes
Paid from	Free	Free
Platforms	api, Linux, macOS, self-hosted, Web, Windows	Linux, self-hosted
Free plan	Yes
Kubernetes support	Yes	Yes
Cloud fault injection	Yes	Yes
Network fault injection	Yes	Yes
Experiment scheduling	Yes
Deployment model	self_hosted	self_hosted
Blast-radius controls	Yes	Yes

Chaos Toolkit vs Krkn-AI

Chaos Toolkit

Krkn-AI