Detect, quarantine, and eliminate flaky tests
Trunk is the CI Reliability Platform designed to eliminate the hidden tax of an unreliable pipeline. By automating flaky test management and providing an intelligent merge queue, we help engineering teams at companies like Zillow, Faire, and Brex save thousands of hours and ship code with confidence.
Trunk provides a flaky test dashboard showing flake frequency, affected PRs, test age, and failure stack traces. AI-powered analysis identifies duplicate failures and generates summaries. Integrations with Slack, Linear, and Jira enable automatic ticket creation and team notifications.
Trunk Flaky Tests detects and quarantines unreliable tests automatically across any language and test runner, preventing false CI failures. Developers see PR comments explaining whether failures are code-related or known flakes, eliminating 10–15 minutes of debugging per incident.
Trunk works with any CI provider, test framework, and build system via CLI integration and JUnit XML/XCResult/Bazel BEP output analysis. No source code access required; setup is provider-agnostic and language-agnostic.
Trunk's merge queue runs parallel testing on non-overlapping code changes, batches up to 100 PRs per run, and auto-bisects failures. Anti-flake protection ensures quarantined test failures do not block merges, enabling sustained 250+ PRs/hour throughput.
By eliminating flaky test re-runs and parallelizing merge queue processing, Trunk reduces CI execution time and compute costs. Customers report 9.4 hours of engineering productivity saved per day and merge times reduced from 30 minutes to 2 minutes.