write-real-e2e-test

Real-backend Playwright tests fail in CI when they rely on fixed delays instead of verified state. You can fix this by using CI-aware polling, confirming backend state transitions, and validating sub-phase completion before asserting gameplay outcomes.

Multiplayer gameplay testing requires isolated browser contexts for each player. Creating separate Playwright contexts ensures multi-user interactions remain deterministic and prevents state bleed when validating complex phase transitions across multiple game iterations.

You write real-backend tests by polling the live backend for verified sub-phase state instead of using hardcoded timing delays. This approach validates actual game phase transitions and keeps multi-iteration scenarios deterministic across different environments.

Hidden race conditions occur when tests assert timing-based state rather than verified backend responses. Stabilizing tests requires waiting on confirmed sub-phase transitions, reusing established helpers, and updating quarantine records to track flaky scenarios.

Playwright can validate sheriff election and night-action specs against a real backend. The tests must use per-player browser isolation and backend state polling to ensure these complex gameplay flows pass deterministically in both local and CI environments.

Debugging flaky real-backend tests involves checking project memory, reusing existing test helpers, and reviewing quarantine records. This helps diagnose timing issues and rewrite assertions to wait for verified backend state rather than arbitrary delays.