Test Adequacy Guard

Version updated for https://github.com/Ostico/test-guard to version v2.0.1.

  • This action is used across all versions by 1 repositories.

Action Type

This is a Composite action.

Go to the GitHub Marketplace to find the latest changes.

Action Summary

Test-Guard automates the task of checking if a pull request has adequate tests. It evaluates every source file in the PR independently, using diff coverage and heuristic test-file matching to determine if changes are tested. The action provides detailed per-file analysis and uses AI-powered evaluation to ensure comprehensive testing coverage.

What’s Changed

A red gate that measured nothing

On a pull_request run whose GITHUB_REF is not refs/pull/<number>/merge, the action aborted before Layer 1 ran:

::error::Could not determine PR number from GITHUB_REF.

Exit 1, no coverage read, no diff examined — but the check goes red, which from the author’s side is indistinguishable from a test-adequacy verdict.

It reproduces on the edited activity type. Editing a PR body produces a fresh run, and that run can arrive with the ref pointing somewhere other than the merge ref. Since GitHub surfaces the latest run per check name, the result was a red required gate on a commit whose earlier run had passed.

The fix

.pull_request.number is present in every pull_request payload, for every activity type. GITHUB_REF carries the number only while GitHub has a merge ref to point the run at.

The action now reads the payload at GITHUB_EVENT_PATH first and keeps GITHUB_REF as the fallback, so every case that worked before still works. An unusable payload — absent, unreadable, malformed, for another event, or with a number that is not an int — yields no number rather than an exception, leaving the fallback its chance. That last case is not hypothetical: true is valid JSON where the number goes, and bool subclasses int, so a naive check would pass True downstream as PR number 1.

The error message, when neither source yields a number, now names both sources and states outright that nothing was measured.

Upgrading

Nothing to change. v2 now points here.

12 tests added (455 in the suite), ruff clean, and each of the four ways to break the change was verified to turn the suite red.

Full detail in #6.