What problem does it solve? Flaky tests that intermittently pass and fail erode trust in CI pipelines, causing teams to ignore genuine failures. This Skill identifies non-deterministic tests from run history, classifies likely root causes, and recommends whether to quarantine or fix each one. ## Core Features & Use Cases - CI Log Parsing: Reads JUnit XML results (GdUnit4, Unity game-ci) and plain-text logs from Godot, Unity, and Unreal test runners to build per-test pass/fail history. - Flakiness Classification: Categorizes tests by failure rate (high, moderate, suspected) and diagnoses likely causes such as timing/async issues, order dependency, random seeds, resource leaks, and floating-point comparisons. - Quarantine Management: Updates the quarantine section of tests/regression-suite.md and optionally writes a full flakiness report to production/qa/, always with explicit user approval. - Use Case: After several CI runs where developers dismiss failures as "probably flaky", run this Skill to statistically confirm which tests are non-deterministic and get engine-specific fix guidance. ## Quick Start Ask the AI to run the test-flakiness skill in scan mode to analyze all available CI logs and report which tests are flaky with recommended actions.
