Weekly CircleCI test-health report posted to Slack

Every Monday at 9:00am, generate a prioritized weekly test-health report for the CircleCI projects I track and deliver it to my team. Run this on a cron schedule. I will give you the list of CircleCI project slugs (in {vcs}/{org}/{repo} form, for example gh/acme/api), the key workflow names for each project, the Slack channel to post to, and the Google Sheet that stores my running trend log.

For each project I track, pull three things from CircleCI. Use Get flaky tests for a project to find tests that passed and failed on the same commit. Use Get test metrics for a workflow, for each important workflow, to surface the slowest tests and the least reliable tests based on the most recent runs. Use Get project summary metrics and trends to read the project's success rate and build-duration metrics along with their trend over the reporting window.

Turn the raw numbers into a ranked report. Rank the worst flaky tests across all my projects, most flaky first, and separately rank the slowest tests by duration. Call out week-over-week regressions: projects or workflows whose success rate dropped or whose median build duration climbed versus last week. Establish last week's baseline from the trend log in my Google Sheet (read the most recent prior row for the previous figures) together with the trend fields CircleCI returns. Because CircleCI's insights window is limited, the Google Sheet is what makes true week-over-week comparison possible over the long run.

Post the report to Slack with Send a Message (Slack Bot) in my chosen channel. Keep the Slack copy short and skimmable: lead with the top offenders, list the handful of worst flaky tests and slowest tests, note any success-rate or build-duration regressions, and end with one plain recommendation on what to fix this week. Use Slack mrkdwn formatting (single asterisks for bold, bullet lines) and keep it tight rather than dumping every metric. Make sure the bot is a member of the channel or has permission to post there.

Then append a one-row summary to the Google Sheet with Append Values so the numbers accumulate into a durable trend log. The row should capture the week (date) plus the headline figures such as overall or per-project success rate, median build duration, the count of flaky tests, the single worst flaky test, and the slowest test. Keep the columns in the same order every week so the sheet builds into something I can chart and compare over time.