How to use Claude to port an Unreal game to iOS
Claude Code helped port an Unreal game to iPhone. Device tests exposed video, build, input and material problems; TestFlight distribution remained unfinished.
koffeecup.net
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Plan a narrow iOS port of this existing Unreal branch. Inspect signing and toolchain logs, reuse existing cutscene and input-state flows, and use my device screenshots to diagnose visual differences. Verify development and distribution builds separately and report unfinished TestFlight checks.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Koffeecup
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Hoomanz booted on iPhone with touch controls and corrected mobile grass materials. TestFlight validation and distribution preparation were still unfinished.
- Human involvement
- The developer reviewed changes, installed builds on an iPhone, wired Blueprint systems and supplied device screenshots and signing feedback.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- The developer supplied the existing branch and reviewed Claude's phased iOS-port plan.
- Claude corrected build, signing and mobile material paths using logs and device screenshots.
- Human play tests caught skip-state errors and guided reuse of the existing Blueprint flow.
Koffeecup used Claude Code to begin an iOS port of Hoomanz from its existing Nintendo branch. The developer asked for a phased plan, then let the agent inspect Unreal configuration, generated Xcode files, signing information and build logs together.
The first device build crashed on its intro video. Removing that video let the work-in-progress game boot. A later build hang came from Apple's asset-catalog tooling; Claude patched the Unreal build toolchain and corrected which rebuilt binaries the automation actually loaded.
Mobile controls exposed another integration error. The agent initially created a new cutscene-skip path that removed UI without restoring gameplay state. The developer directed it back to the existing Blueprint flow. A device screenshot also helped diagnose bright cyan grass as an invalid mobile virtual-texture path, leading to a separate material branch for iOS.
The result was a game running on an iPhone with touch controls and improved visuals. Distribution was still unfinished: the temporary asset-catalog workaround later caused TestFlight icon validation problems. Human device trials, screenshots and code-flow review were the checkpoints that separated a plausible port from behavior that actually matched the game.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Studio devlog · Koffeecup
Frequently asked questions
What first prevented the game from booting?
The intro video's device path caused an immediate crash.
Why did the first skip button fail?
It bypassed the existing flow that restored gameplay state.
Was distribution complete?
No. TestFlight validation and later production phases still required work.
More details
About this account
The developer supplied the existing branch and reviewed Claude's phased iOS-port plan. Claude corrected build, signing and mobile material paths using logs and device screenshots.
Original studio account opened. The game booted with mobile controls and corrected visuals; distribution/TestFlight signing and later production phases remained work in progress. Suggested prompt written by Agents In Use; no original task prompt was published in the verified account.
Credit: Koffeecup — Studio devlog
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Unreal Engine · Xcode · iPhone · Blueprints
Game porting · Device screenshots
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.