LitmusChaos is a chaos engineering platform for Kubernetes. It provides a catalog of pre-built fault injections — pod delete, network latency, DNS chaos, disk fill, CPU hog, node drain, and cloud-provider-specific faults for AWS, Azure, and GCP — packaged as reusable experiments.
Experiments are defined as ChaosExperiment and ChaosEngine custom resources and executed by the chaos-operator, which launches a runner pod that in turn invokes the actual fault logic (often via Ansible or Go-based executors). Workflows chain experiments together and are authored in Argo Workflows under the hood, so Litmus inherits Argo’s DAG semantics. The ChaosCenter web console provides multi-cluster scheduling, RBAC, and probes that assert application health before, during, and after an experiment — a run fails if the probes detect regressions. ChaosHub is a public catalog of community-contributed experiments packaged as charts.
Litmus is a CNCF incubating project maintained by Harness, and it competes with Chaos Mesh, Gremlin, and AWS Fault Injection Simulator. Its distinguishing traits are the pluggable probe model and the use of Argo Workflows for orchestration.