// the find
mikepenz/action-junit-report
Reports junit test results as GitHub Pull Request Check
A GitHub Action that parses JUnit XML reports and posts results as PR check annotations, job summaries, and PR comments. It handles most JUnit variants (Maven Surefire, pytest, Catch2, xUnit, nextest, etc.) without requiring any language-specific setup beyond producing XML. Useful for any team already generating JUnit output who wants visible test failures on PRs without writing their own glue code.
The flaky test detection and retry-awareness is genuinely useful — if your test runner re-runs failures, the action can ignore the original failure and only surface tests that failed every attempt. The `workflow_run` pattern for forked PRs is handled correctly with a documented split-workflow example, which is the actual right answer to the fork permission problem rather than the common wrong one. Output variables (passed, failed, total, etc.) let downstream steps act on results without parsing HTML. The breadth of test fixture XML files in the repo suggests real-world compatibility testing against edge cases, not just happy-path coverage.
The heap OOM warning in the README is a red flag — a JUnit parser shouldn't need 4GB of Node heap; that points to loading all test results into memory rather than streaming. The `annotations_limit` option exists, which means large test suites hit GitHub API limits and the workaround is to just stop annotating, not to paginate gracefully. The `update_check` flag that switches API modes for >50 annotations feels like a leaky abstraction the user shouldn't need to know about. Configuration surface is enormous (40+ inputs) for what is fundamentally a file-parse-and-post operation — every edge case got its own flag instead of sensible defaults that cover 90% of cases.