Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Package and ship an Unreal Engine 5 project: the Platforms menu Package Project flow, build configurations (Development vs Shipping), cooking content, packaging settings and the Game Default Map, and command-line builds with RunUAT BuildCookRun. Use when packaging a build, making a shipping build, cooking content, configuring packaging settings, or when the user mentions package Unreal, cook content, shipping build, or BuildCookRun.
.claude/skills/gamedev-skills-unreal-packaging/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -15% | 0% |
Turn a UE5 project into a runnable, distributable build: choose the right build configuration, cook content, set the launch map, and package — from the editor or the command line. Targets UE 5.8.
content, configuring Packaging / Maps & Modes settings, or automating builds with RunUAT BuildCookRun for CI.
*.uproject and Config/Default*.ini, and the goal is a packagedplayer rather than running in-editor.
When not to use: storefront submission/release process → steam-publish / itch-publish. Editor-time gameplay/iteration is not packaging.
packaged build loads first. Wrong/empty here is the most common "packaged game is black" cause.
stats/console for testing) vs Shipping (all optimizations, debugging tools stripped — for release). DebugGame/Debug are for debugging engine/game code and aren't for distribution; DebugGame isn't available for Blueprint-only projects.
and packs them into .pak files. Packaging bundles the compiled executable + cooked content into a standalone, distributable set of files. Packaging runs a cook as part of it.
set the Binary Configuration → Package Project → pick an output folder.
RunUAT BuildCookRun)for repeatable/CI builds.
cook, full-rebuild, compression, and whether to build all maps.
executable and confirm it loads the right map and runs.
textPlatforms (toolbar) -> Windows -> Binary Configuration -> Development | Shipping -> Content Management -> Package Project -> choose/confirm the staging output folder
bash# Cook + build + stage + pak + archive a Shipping Windows build. RunUAT BuildCookRun \ -project="C:/Path/MyGame.uproject" \ -noP4 -platform=Win64 -clientconfig=Shipping \ -cook -allmaps -build -stage -pak -archive \ -archivedirectory="C:/Builds/MyGame"
RunUAT lives in Engine/Build/BatchFiles/ (RunUAT.bat on Windows, RunUAT.sh on macOS/Linux). Drop -allmaps and pass -map=Map1+Map2 to cook a subset.
bashRunUAT BuildCookRun -project="C:/Path/MyGame.uproject" -noP4 \ -platform=Win64 -clientconfig=Development -cook -skipstage
cooked). Set it in Maps & Modes and ensure it's included in the cook.
Shipping. Conversely, Shipping strips UE_LOG/console, so debugging a Shipping-only issue needs Development or Test.
settings' maps/directories to cook, or cook all maps.
tools, Android SDK/NDK, console SDKs). Install the platform's prerequisites.
performance. Profile and move hot logic to C++ (unreal-cpp-gameplay) instead.
incremental. Don't mistake a slow first cook for a hang.
(https://dev.epicgames.com/documentation/en-us/unreal-engine/packaging-your-project) and the Build Configurations / BuildCookRun references.
steam-publish / itch-publish — distributing the packaged build to a storefront.unreal-cpp-gameplay — moving hot logic to C++ now that BP nativization is gone.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→pass | 24,263 | 21,386 | -12% | 1 | 1 | 0% | 3,305 | 4,091 | +24% | 0 | 0 | — |
case-05 | pass→pass | 19,821 | 8,868 | -55% | 1 | 1 | 0% | 2,240 | 2,607 | +16% | 0 | 0 | — |
case-03 | pass→pass | 14,906 | 10,744 | -28% | 1 | 1 | 0% | 2,569 | 3,131 | +22% | 0 | 0 | — |
case-01 | fail→pass | 19,879 | 17,853 | -10% | 1 | 1 | 0% | 2,406 | 3,397 | +41% | 0 | 0 | — |
case-02 | fail→fail | 18,838 | 20,265 | +8% | 1 | 1 | 0% | 3,475 | 4,105 | +18% | 0 | 0 | — |
case-06 | pass→pass | 19,151 | 10,848 | -43% | 1 | 1 | 0% | 2,390 | 3,102 | +30% | 0 | 0 | — |
case-07 | pass→pass | 15,042 | 6,505 | -57% | 1 | 1 | 0% | 1,689 | 2,418 | +43% | 0 | 0 | — |
case-08 | fail→pass | 10,699 | 13,707 | +28% | 1 | 1 | 0% | 1,741 | 2,201 | +26% | 0 | 0 | — |
case-09 | pass→pass | 10,226 | 7,718 | -25% | 1 | 1 | 0% | 1,478 | 2,484 | +68% | 0 | 0 | — |
case-10 | pass→pass | 19,913 | 13,859 | -30% | 1 | 1 | 0% | 1,789 | 2,658 | +49% | 0 | 0 | — |
case-11 | pass→pass | 12,062 | 10,278 | -15% | 1 | 1 | 0% | 1,986 | 2,939 | +48% | 0 | 0 | — |
case-12 | pass→pass | 11,596 | 8,335 | -28% | 1 | 1 | 0% | 1,930 | 2,651 | +37% | 0 | 0 | — |
case-13 | pass→pass | 11,257 | 8,119 | -28% | 1 | 1 | 0% | 1,846 | 2,549 | +38% | 0 | 0 | — |
case-14 | fail→pass | 20,432 | 12,098 | -41% | 1 | 1 | 0% | 2,441 | 2,973 | +22% | 0 | 0 | — |
case-15 | pass→pass | 14,227 | 8,693 | -39% | 1 | 1 | 0% | 2,231 | 2,533 | +14% | 0 | 0 | — |
case-16 | fail→pass | 7,959 | 3,675 | -54% | 1 | 1 | 0% | 1,178 | 1,745 | +48% | 0 | 0 | — |
case-22 | pass→pass | 13,293 | 8,812 | -34% | 1 | 1 | 0% | 1,903 | 2,632 | +38% | 0 | 0 | — |
case-17 | pass→pass | 4,937 | 3,786 | -23% | 1 | 1 | 0% | 811 | 1,852 | +128% | 0 | 0 | — |
case-18 | pass→fail | 14,912 | 11,906 | -20% | 1 | 1 | 0% | 2,092 | 2,838 | +36% | 0 | 0 | — |
case-19 | pass→pass | 12,669 | 9,021 | -29% | 1 | 1 | 0% | 1,934 | 2,802 | +45% | 0 | 0 | — |
case-20 | pass→pass | 14,081 | 10,884 | -23% | 1 | 1 | 0% | 2,205 | 2,740 | +24% | 0 | 0 | — |
case-21 | fail→pass | 10,813 | 1,877 | -83% | 1 | 1 | 0% | 1,641 | 1,389 | -15% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.