日本語版

CI Flaky Analyzer

The CLI continuously ingests JUnit-formatted test results, scores flaky behavior, and visualizes it. It ships with npm scripts for CI integration and automates weekly reporting.

Highlights

• Separates CLI subcommands such as flaky parse, flaky analyze, and flaky issue to cover collection through reporting.
• Streams large JUnit XML files and persists analysis as JSONL and HTML.
• The weekly command updates docs/weekly-summary.md to keep knowledge sharing automatic.

Key Artifacts

• README.md — Setup instructions and command list.
• config/flaky.yml — Scoring and window configuration.
• demo/ — Sample JUnit logs and HTML report inputs.
• out/index.html — Visualization of the analysis results.

How to Reproduce

1. Run npm install inside projects/03-ci-flaky/.
2. For demo logs, run npm run demo:parse → npm run demo:analyze and inspect the generated out/ directory.
3. In production, feed CI JUnit XML files and wire npm run ci:analyze or npm run ci:issue into the workflow.

Next Steps

• Connect to Slack webhooks or the GitHub Issues API to take flaky issue from dry-run to production ticketing.
• Export the CSV output (out/summary.csv) to BI tools for long-term trend analysis.
• See the weekly summary list for automation updates.