// the take
A chaos-engineering library you require into an Express/Node app to inject process-level failures — crashes, memory/CPU spikes, uncaught exceptions, unhandled rejections, forced 500s — and trigger them via a REST API, CLI, or a bundled React dashboard. It's aimed at teams who want to poke at Node-specific failure modes before going to production, rather than at infra-level chaos (network partitions, pod kills).
Targets failure modes other chaos tools skip entirely — unhandled promise rejections, uncaught exceptions, blocked event loop — instead of just killing containers or cutting network links. Zero-code smoke-test path (`node -r node-chaos-monkey app.js`) lets you trigger random chaos without touching app code. Ships a real UI and a websocket feed of prank activity, not just a CLI, so you can watch what happened and whether your monitoring caught it. Pluggable schedule model (immediate, one-time, peaks) separates 'what breaks' from 'when it breaks', making it straightforward to add new pranks.
Explicitly alpha and now unmaintained — last push is March 2023, README still carries the 'Alpha mode, stabilizing' banner, and there's no evidence of testing against modern Node LTS versions. The chaos endpoints (`/chaos`, `POST /chaos/random`) are unauthenticated HTTP routes with no built-in restriction to non-prod environments — anyone who can reach that port can crash your process or force errors on live traffic if this ships by accident. Only 9 pranks and 3 schedule types, with nothing for network-level chaos (latency, partitions, dependency failures), so you'll still need a separate tool like Toxiproxy for anything beyond in-process faults. With the project dormant, any gaps or bugs you hit are yours to fix.
We read the README, the directory tree and a key source file in October 2026. Nobody pays to be listed. How we judge →