ue-unit-test

Run Sentry unit tests via Unreal Engine automation framework.

Prerequisite: the project must be built first (a compile-only editor build is sufficient). If unsure whether it's been built, ask the user. If not built, suggest running /ue-build first.

1. Check the UNREAL_ENGINE_ROOT environment variable (use PowerShell .NET method on Windows: [System.Environment]::GetEnvironmentVariable('UNREAL_ENGINE_ROOT')). If unset or the path doesn't exist, ask the user for the engine path.

2. Determine the engine version by reading the EngineAssociation field in sample/SentryPlayground.uproject (e.g., "EngineAssociation": "5.7" - UE5, "EngineAssociation": "4.27" - UE4). Select the headless editor binary based on host OS and engine version:

Host OS	UE5 binary path	UE4 binary path
Windows	Engine/Binaries/Win64/UnrealEditor-Cmd.exe	Engine/Binaries/Win64/UE4Editor-Cmd.exe
macOS	Engine/Binaries/Mac/UnrealEditor-Cmd	Engine/Binaries/Mac/UE4Editor-Cmd
Linux	Engine/Binaries/Linux/UnrealEditor-Cmd	Engine/Binaries/Linux/UE4Editor-Cmd

3. Resolve the repo root and run the automation command:

REPO=$(git rev-parse --show-toplevel)

# Windows UE5
"$UNREAL_ENGINE_ROOT/Engine/Binaries/Win64/UnrealEditor-Cmd.exe" \
    "$REPO/sample/SentryPlayground.uproject" \
    -ReportExportPath="$REPO/sample/Saved/Automation" \
    -ExecCmds="Automation RunTests Sentry;quit" \
    -TestExit="Automation Test Queue Empty" \
    -Unattended -NoPause -NoSplash -NullRHI

# macOS UE5
"$UNREAL_ENGINE_ROOT/Engine/Binaries/Mac/UnrealEditor-Cmd" \
    "$REPO/sample/SentryPlayground.uproject" \
    -ReportExportPath="$REPO/sample/Saved/Automation" \
    -ExecCmds="Automation RunTests Sentry;quit" \
    -TestExit="Automation Test Queue Empty" \
    -Unattended -NoPause -NoSplash -NullRHI

# Linux UE5
"$UNREAL_ENGINE_ROOT/Engine/Binaries/Linux/UnrealEditor-Cmd" \
    "$REPO/sample/SentryPlayground.uproject" \
    -ReportExportPath="$REPO/sample/Saved/Automation" \
    -ExecCmds="Automation RunTests Sentry;quit" \
    -TestExit="Automation Test Queue Empty" \
    -Unattended -NoPause -NoSplash -NullRHI

The test run can take longer than the default command timeout — run it in the background and monitor its output, or raise the timeout accordingly.

4. Report the results. The editor's exit code is not a reliable pass/fail signal for automation runs — read the report at sample/Saved/Automation/index.json instead:

   • Summarize the number of passed and failed tests (succeeded / failed counts).
   • For each failed test, report its full name and the error messages from its report entries.

   If the report is missing or the run crashed, check sample/Saved/Logs/ for the root cause.