القائمة
Compare test coverage on the current branch against main and fail if it regressed. Run before opening any PR — CI enforces a coverage regression gate, and this catches it locally first.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
LSP-first code navigation for ZeeKayDa.Auth — load the LSP tool up front and use it for all symbol-level lookups. Applies to every agent that reads or writes C# in this repository.
Verify and fix code formatting with dotnet format. Run before opening any PR, and whenever formatting errors are reported by CI or the Stop hook.
Housekeeping after a PR merges — delete the local branch, pull main, remove resolved blockers from draft PRs and issues, and check whether the parent epic can close. Run whenever a PR has been merged.
Write a well-structured GitHub issue for ZeeKayDa.Auth — decides between an ADR issue and an implementation issue, applies the issue templates, hierarchy, and labels, and links sub-issues to their epic. Use when fleshing out a new idea, filing a bug, creating design or implementation issues, or triaging incoming issues.
Set up a git worktree for reviewing or auditing a branch or PR so the review runs against the right code. Use before any code review, audit, or acceptance-criteria verification that targets a branch other than the currently checked-out one.
Security review checklist for ZeeKayDa.Auth. Apply to every change touching tokens, cryptography, endpoints, or storage — during security review, and as a developer self-check before opening a PR that touches those areas.
| name | check-code-coverage |
| description | Compare test coverage on the current branch against main and fail if it regressed. Run before opening any PR — CI enforces a coverage regression gate, and this catches it locally first. |
| allowed-tools | ["Bash(dotnet *)","Bash(git worktree *)","Bash(cd *)"] |
To verify that code coverage hasn't dropped too far, do the following
Measuring the coverage in the current branch can be done by running:
dotnet restore
dotnet build --no-restore --configuration Release
dotnet test tests/ZeeKayDa.Auth.Tests/ \
--no-build --configuration Release \
--collect:"XPlat Code Coverage" --results-directory ./TestResults/pr \
-- DataCollectionRunSettings.DataCollectors.DataCollector.Configuration.Include="[ZeeKayDa.Auth]*"
dotnet test tests/ZeeKayDa.Auth.AspNetCore.Tests/ \
--no-build --configuration Release \
--collect:"XPlat Code Coverage" --results-directory ./TestResults/pr \
-- DataCollectionRunSettings.DataCollectors.DataCollector.Configuration.Include="[ZeeKayDa.Auth.AspNetCore]*"
dotnet test tests/ZeeKayDa.Auth.AzureKeyVault.Tests/ \
--no-build --configuration Release \
--collect:"XPlat Code Coverage" --results-directory ./TestResults/pr \
-- DataCollectionRunSettings.DataCollectors.DataCollector.Configuration.Include="[ZeeKayDa.Auth.AzureKeyVault]*"
dotnet test tests/ZeeKayDa.Auth.FileSystem.Tests/ \
--no-build --configuration Release \
--collect:"XPlat Code Coverage" --results-directory ./TestResults/pr \
-- DataCollectionRunSettings.DataCollectors.DataCollector.Configuration.Include="[ZeeKayDa.Auth.FileSystem]*"
Rather than checking out and rebuilding main in a worktree, download the coverage-Linux artifact from main's most recent successful CI run (the same one coverage-regression itself now uses — see .github/workflows/ci.yml). This requires the gh CLI to be authenticated:
gh run download --repo ChrisKlug/zeekayda-auth \
-n coverage-Linux \
-D ./TestResults/base \
$(gh run list --repo ChrisKlug/zeekayda-auth --workflow ci.yml --branch main --event push --status success --limit 1 --json databaseId --jq '.[0].databaseId')
To compare the results, you can run the check_coverage_regression.cs file like this
dotnet run .github/scripts/check_coverage_regression.cs -- ./TestResults/pr ./TestResults/base
To see if there are any formatting issues
If the execution returns a 0 exit code, coverage is good enough. If it returns a non-0 result, it is not.
If the coverage is not good enough, the output from the execution will contain the result of the check. This should be possible to use to figure out what tests are missing.
Once the check has been performed, remove the downloaded artifact:
rm -rf ./TestResults