vladyslavdmitriiev/playwright-mutation-gate

This is a testing tool for teams that write end to end tests with Playwright, the browser automation framework. Its purpose is to catch a specific problem: tests that pass but do not actually check anything meaningful, sometimes called hollow assertions. The idea is simple. You mark the main assertion in each test with a comment, then run this tool as part of your CI pipeline. It takes that marked assertion, flips it to expect the opposite result, and reruns the test in a temporary copy. If the test still passes after the assertion was reversed, that means the original assertion was never really testing anything, and the tool fails the build to flag it. If the test fails once reversed, that is a good sign, since it shows the assertion was actually connected to real behavior. The tool reports each test as killed, meaning the assertion is doing real work, survived, meaning it is not, or cannot mutate, meaning the assertion could not be safely reversed. By default, any test without a marker also fails the check, though this can be turned off while a team is still adding markers to an older test suite. There is also a system for deliberately skipped tests, where a comment can record an owner and an expiration date so the exception does not linger forever unnoticed. A newer, optional mode goes further by rewriting the actual data a test's assertion depends on as it travels over the network, rather than just flipping the assertion itself. This checks whether an assertion is truly tied to the data it claims to verify, rather than a cached or coincidentally correct value. It ships as an npm package, requires Node version 20 or higher plus an existing Playwright setup, and includes a sample GitHub Actions workflow for running it in CI. Cost scales with the number of marked assertions, so the README suggests running it only on changed test files per pull request, or as a full nightly sweep instead of on every commit.