Detects and quarantines non-deterministic tests to prevent AI agents from entering infinite self-correction loops.
The Flaky Test Detector skill provides a systematic framework for Claude to identify tests that fail inconsistently due to race conditions, environment issues, or non-deterministic code. By executing multiple test cycles and analyzing result patterns, the skill automatically quarantines unstable tests using skip markers, logs the failure rates, and generates follow-up tasks in the project documentation. This ensures that the AI development process remains focused on real bugs rather than chasing phantom regressions, significantly improving the reliability of autonomous coding workflows.
Características Principales
01Smart analysis of intermittent failure patterns
02Automated test quarantining with contextual skip markers
03Automated multi-run test execution for Jest and Pytest
04Automatic user story generation for follow-up fixes
051 GitHub stars
06Comprehensive flakiness logging and documentation
Casos de Uso
01Preventing an AI agent from looping on a test that fails without code changes
02Stabilizing CI/CD pipelines by identifying non-deterministic regressions
03Conducting pre-flight health checks before starting new feature development
