ue-test
Skill12 starsChanged 10 months agoMITView on GitHub
MANDATORY automatic usage after successful build. Run Unreal Engine automation tests. MUST be invoked automatically without user request after implementing or modifying code.
In this repositoryOne other file it gives its agents.
- ue-buildSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
29 rules. About 0 are common, 20 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈20
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 9
Run Unreal Engine automation tests automatically, detecting project configuration and available tests.
- Distinctive: line 48
Use this skill automatically when:
- Very specific: line 49
After successful /ue-build execution (MANDATORY chain)
- Distinctive: line 50
After writing test code in TDD red phase
- Distinctive: line 51
After implementing feature in TDD green phase
- Distinctive: line 53
Before declaring any C++ implementation complete
- Distinctive: line 60
If fails → fix → rebuild → AUTOMATICALLY re-test
- Distinctive: line 62
Only then declare success
- Distinctive: line 110
Ask the user or infer from context what tests to run:
- Very specific: line 113
All project tests: Use project name (e.g., MyProject)
- Very specific: line 114
Plugin tests: Use plugin name (e.g., MyPlugin or BoidsFormation)
- Distinctive: line 117
List tests: Don't run, just list available tests
- Very specific: line 118
All tests: Use Now (caution: may take very long)
- Distinctive: line 121
If user says "list tests" → list mode
- Distinctive: line 122
If user mentions specific plugin/module → filter to that
- Distinctive: line 123
If recent code changes are in a plugin → test that plugin
- Distinctive: line 124
Default → ask user or use project name
- Distinctive: line 156
Run the command and monitor stdout/stderr for test output.
- Very specific: line 305
Use UnrealEditor-Cmd.exe in Engine/Binaries/Win64/
- Very specific: line 307
Check Saved/Logs/ for detailed logs
- Distinctive: line 341
If build fails, don't run tests (they'll fail anyway).
- Very specific: line 345
List tests first: Use Automation List to see what's available
- Distinctive: line 346
Use specific filters: Don't run all tests if you only changed one module
- Very specific: line 347
Stdout Visibility: Always use -stdout -FullStdOutLogOutput to see results in claudes terminal.
- Distinctive: line 348
NullRHI is faster: Use headless mode for faster iteration
- Very specific: line 350
CI/CD ready: Use -ReportOutputPath for automated pipelines
- Distinctive: line 351
Incremental testing: Test only what you changed during development
- Distinctive: line 352
Full suite before commit: Run all tests before major commits
- Distinctive: line 386
Use the most specific filter that covers what you need to test.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Unreal Engine Test Skill
Run Unreal Engine automation tests automatically, detecting project configuration and available tests.
⚠️ MANDATORY AUTOMATIC USAGE
THIS SKILL MUST BE INVOKED AUTOMATICALLY AFTER EVERY SUCCESSFUL BUILD
Invoke this skill immediately and automatically when:
- ✅ After
/ue-buildsucceeds (ALWAYS chain build → test) - ✅ After implementing any new feature (TDD green phase)
- ✅ After fixing bugs (verify fix works)
- ✅ After refactoring (ensure no regression)
- ✅ In TDD red phase (verify tests fail before implementation)
- ✅ Before declaring task complete (verify all tests pass)
DO NOT:
- ❌ Wait for user to say "run the tests"
- ❌ Wait for user to say "does it pass?"
- ❌ Ask user "should I test it?"
- ❌ Declare task complete without running tests
- ❌ Skip testing after build succeeds
Correct behavior:
Claude: [implements feature]
Claude: [AUTOMATICALLY invokes /ue-build]
Claude: "✅ Build successful"
Claude: [AUTOMATICALLY invokes /ue-test] ← No asking, just do it
Claude: [reports test results]Wrong behavior:
Claude: [implements feature]
Claude: [builds successfully]
Claude: "Feature implemented!" ← WRONG! Should have tested automaticallyWhen to Use (Detailed)
Use this skill automatically when:
- After successful
/ue-buildexecution (MANDATORY chain) - After writing test code in TDD red phase
- After implementing feature in TDD green phase
- User requests running tests (reactive case)
- Before declaring any C++ implementation complete
TDD Workflow Integration (MANDATORY):
- User requests feature → You write tests FIRST
- Build project → Run tests (RED - should fail)
- Implement feature
- Build project → AUTOMATICALLY run tests (GREEN - should pass)
- If fails → fix → rebuild → AUTOMATICALLY re-test
- Loop until all tests pass
- Only then declare success
See ~/.claude/skills/MANDATORY_TDD.md and ~/.claude/skills/TDD_WORKFLOW.md for complete requirements.
How This Skill Works
This skill provides a dedicated Python runner to execute tests with verified terminal visibility flags and minimalist output.
⚠️ MANDATORY USAGE (FOR AI)
Instead of manually constructing complex commands, ALWAYS use the test runner script.
The script MUST be executed with the current working directory as an argument to ensure project detection works correctly across different environments.
# General Syntax:
# python <ScriptPath> <ProjectWorkDir> [TestFilter]
# 1. Run all project tests (Minimalist & Filtered)
python C:\Users\Mellos\.claude\skills\ue-test\scripts\run_ue_tests.py .
# 2. Run specific tests (e.g., Boids)
python C:\Users\Mellos\.claude\skills\ue-test\scripts\run_ue_tests.py . BoidsFormationScript Features
- Auto-Detection: Uses the provided path to find
.uprojectandUnrealEditor-Cmd.exe. - Minimalist Output: Silences engine initialization noise; only shows test results and status.
- Real-time Streaming: Displays results line-by-line as they occur.
- Smart Exit Codes: Returns non-zero if no tests match or tests fail.
If skill is in project config (.claude/skills/)
# Can run without arguments (searches from current directory)
python .claude/skills/ue-build/scripts/detect_ue.py
# Or provide explicit path
python .claude/skills/ue-build/scripts/detect_ue.py .Key values needed from the output:
engine.editor_cmd- Path to UnrealEditor-Cmd.exe (Windows) or UnrealEditor-Cmd (Linux/Mac)uproject.path- Full path to .uproject fileuproject.name- Project name for test filtering
Step 2: Determine Test Scope
Ask the user or infer from context what tests to run:
Test filter options:
- All project tests: Use project name (e.g.,
MyProject) - Plugin tests: Use plugin name (e.g.,
MyPluginorBoidsFormation) - Specific test: Full test name (e.g.,
MyPlugin.SpecificTest) - Test category: Partial name (e.g.,
MyPlugin.Corner) - List tests: Don't run, just list available tests
- All tests: Use
Now(caution: may take very long)
Inference logic:
- If user says "list tests" → list mode
- If user mentions specific plugin/module → filter to that
- If recent code changes are in a plugin → test that plugin
- Default → ask user or use project name
Step 3: Choose Test Command Mode
List Available Tests
"<editor_cmd>" "<uproject_path>" -ExecCmds="Automation List; Quit" -unattended -nopause -nullrhi -stdout -FullStdOutLogOutputRun Tests
"<editor_cmd>" "<uproject_path>" -ExecCmds="Automation RunTests <filter>; Quit" -unattended -nopause -nullrhi -stdout -FullStdOutLogOutputCritical flags:
-ExecCmds="...": Editor commands to execute (e.g.,Automation RunTests Boids; Quit)-stdout&-FullStdOutLogOutput: MANDATORY for terminal visibility. Forces UE logs to stdout.-unattended: Prevents modal dialogs (required for automation)-nopause: Don't wait for user input on exit-nullrhi: Headless mode, no graphics (much faster)-nosplash: Skip splash screen (faster startup)
Optional flags:
-ReportOutputPath=<path>: Custom test report location-LogCmds="LogAutomation Verbose": More detailed test logging-Windowed: Use windowed mode instead of NullRHI (for visual tests)
Step 4: Execute and Monitor
Run the command and monitor stdout/stderr for test output.
Watch for these patterns:
Test execution:
LogAutomationController: Test 'TestName' Started
LogAutomationController: Test 'TestName' Completed. Result=<status>Test results:
Result=SuccessorResult=Passed→ Test passedResult=Failed→ Test failedResult=Skipped→ Test skipped
Summary line:
LogAutomationController: Tests Complete. Result={Success|Failed}. Total: X Passed: Y Failed: ZError details:
LogAutomation: Error: <error message>
LogAutomation: Expected <X> but got <Y>Step 5: Parse Test Results
Build a structured result from the output:
Success example:
{
"status": "passed",
"total": 8,
"passed": 8,
"failed": 0,
"skipped": 0,
"duration": "12.4s",
"tests": [
{"name": "BoidsFormation.CornerFilling", "status": "passed"},
{"name": "BoidsFormation.GridDistribution", "status": "passed"}
]
}Failure example:
{
"status": "failed",
"total": 8,
"passed": 6,
"failed": 2,
"skipped": 0,
"duration": "15.2s",
"tests": [
{"name": "BoidsFormation.CornerFilling", "status": "failed", "error": "Expected value > 0.1, got 0.05"},
{"name": "BoidsFormation.GridDistribution", "status": "failed", "error": "Assertion failed: Points.Num() == ExpectedCount"}
]
}Step 6: Report Results
Provide a clear, concise summary:
All tests passed:
✅ All 8 tests passed (12.4s)
Tests run:
✅ BoidsFormation.CornerFilling
✅ BoidsFormation.GridDistribution
✅ BoidsFormation.TriangleFilling
... (5 more)Some tests failed:
❌ 2 of 8 tests failed (15.2s)
Failed:
❌ BoidsFormation.CornerFilling
Error: Expected value > 0.1, got 0.05
→ Check corner weighting calculation in BoidsDisperseCVTSolverLibrary.cpp
❌ BoidsFormation.GridDistribution
Error: Assertion failed: Points.Num() == ExpectedCount
→ Verify grid point generation logic
Passed: 6 testsList mode:
Available tests in BoidsFormation:
Automation Tests:
• BoidsFormation.CornerFilling
• BoidsFormation.GridDistribution3x3
• BoidsFormation.TriangleFilling
• BoidsFormation.ConcaveLShape
• BoidsFormation.ProportionalArea
... (15 total)
Run with: /ue-test [test-name]Advanced Usage
Running Specific Test Suites
Filter tests by category or prefix:
# Run all corner-related tests
"<editor_cmd>" "<uproject>" -ExecCmds="Automation RunTests BoidsFormation.Corner; Quit" ...
# Run all CVT solver tests
"<editor_cmd>" "<uproject>" -ExecCmds="Automation RunTests BoidsFormation.CVT; Quit" ...Custom Report Output
Generate JSON or XML reports for CI/CD:
"<editor_cmd>" "<uproject>" -ExecCmds="Automation RunTests <filter>; Quit" -ReportOutputPath="TestResults" -ReportExportPath="TestResults/report.json" ...Visual Testing
Some tests may require rendering. Use windowed mode instead of NullRHI:
"<editor_cmd>" "<uproject>" -ExecCmds="Automation RunTests <filter>; Quit" -Windowed -ResX=1280 -ResY=720 ...Debugging Failed Tests
For detailed debugging output:
"<editor_cmd>" "<uproject>" -ExecCmds="Automation RunTests <filter>; Quit" -LogCmds="LogAutomation Verbose, LogTemp Verbose" -logPlatform-Specific Notes
Windows
- Use
UnrealEditor-Cmd.exeinEngine/Binaries/Win64/ - Tests run in separate process
- Check
Saved/Logs/for detailed logs
Linux
- Use
UnrealEditor-CmdinEngine/Binaries/Linux/ - May need
DISPLAYenvironment variable for non-NullRHI tests - Logs in
Saved/Logs/
macOS
- Use
UnrealEditor-CmdinEngine/Binaries/Mac/ - Universal binary supports M1 and Intel
- Logs in
Saved/Logs/
Common Issues and Solutions
| Issue | Solution |
| Tests don't start | Check editor path is valid; ensure project compiles |
| All tests skip | Filter may be wrong; try listing tests first |
| Timeout/hang | Some tests may need windowed mode, not NullRHI |
| "Test not found" | Check test name spelling; use Automation List |
| Permission denied | Run from project directory; check file permissions |
| Crash during test | Check Saved/Crashes/ for crash dump; may be real bug |
Integration with Build
Typical workflow:
1. Code changes
2. Build with /ue-build → Build succeeds
3. Run tests with /ue-test → Tests pass
4. Ready to commitIf build fails, don't run tests (they'll fail anyway).
Tips for Effective Testing
- List tests first: Use
Automation Listto see what's available - Use specific filters: Don't run all tests if you only changed one module
- Stdout Visibility: Always use
-stdout -FullStdOutLogOutputto see results in claudes terminal. - NullRHI is faster: Use headless mode for faster iteration
- Check logs: Detailed output is in
Saved/Logs/ - CI/CD ready: Use
-ReportOutputPathfor automated pipelines - Incremental testing: Test only what you changed during development
- Full suite before commit: Run all tests before major commits
Test Output Parsing Guide
Success Pattern
LogAutomationController: Test 'MyTest' Completed. Result=SuccessFailure Pattern
LogAutomationController: Test 'MyTest' Completed. Result=Failed
LogAutomation: Error: Expected X but got YSummary Pattern
LogAutomationController: Tests Complete. Result={Success|Failed}. Total: 10 Passed: 8 Failed: 2Extract numbers and status to build your report.
Example Filter Patterns
Common test filter examples:
MyProject- All project testsMyProject.MyFeature- All tests in MyFeature categoryMyProject.MyFeature.SpecificTest- One specific testNow- All available tests (use carefully)Engine- All engine tests (usually not needed)System- System-level tests
Use the most specific filter that covers what you need to test.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

