Over the past few weeks, I’ve been steadily developing a MOBA-style template in Unreal Engine—and today, I’m excited to say that the free version is complete and available for download!
But this wasn’t built the traditional way. I relied heavily on prompt engineering and what I now understand is called context engineering to create a development workflow that leveraged AI to generate functional Python scripts, modular Blueprints, and in-world level design tools.
Let me show you exactly how I made it happen.
Prompt Engineering is the process of crafting intentional, specific questions or instructions that get the AI to return useful results. But to really go deep with this process, I needed more than one-off prompts.
I needed Context Engineering.
In a traditional dev workflow, you rely on your editor and docs to "remember" what you’ve done.
With AI? You need to feed the model your context. What Blueprints are you using? What variables exist? What do your actors look like?
Once I gave ChatGPT the complete structure of my project (from Blueprint names to Enum options to the specific actor roles), it was able to:
Understand how the actors interacted
Track TeamID systems
Generate valid Python scripts that worked with my custom dummy Blueprints
Help me modularize attack logic with interfaces and components
This meant that I wasn’t just asking it to "write a Python script"—I was showing it how my world worked.
Here’s a quick overview of the prompt-driven milestones that brought the project to life:
1. Spawning Lanes, Turrets, Minions, and the Player
I created dummy Blueprints like BP_LaneBuilder, BP_TurretSpawnPoint, and BP_PlayerStartMarker to act as placeholders.
From there, I used prompts like:
"Write a Python script that finds all BP_LaneBuilder actors and spawns a BP_Lane at their location, with spline length 7000 and 4 segments."
Or:
"Use the TeamID enum to assign Team1 to the PlayerCharacter and Team2 to the spawned minions."
These prompts returned working Python scripts I could copy directly into the Unreal Python editor.
2. Team ID + Interface Checks
I prompted:
"BP_Minion and BP_Turret should not target actors unless CanBeTargeted() returns true, and TeamIDs do not match."
From that, the AI helped write BPI_TeamIdentifiable, implemented clean checks in each actor, and helped set up target filtering logic across AI perception systems.
3. AI Attack Behavior
I prompted for behavior like:
"When a minion detects a player, it should stop moving and call DoAttack() via an interface if alive."
This eventually evolved into a clean interface-based system with custom Blueprint components like BPC_Attack.
What made this workflow powerful wasn’t just the prompting—it was the ongoing feedback loop.
If the AI misunderstood, I corrected it with specifics: “That variable is actually called lane_target, not Lane.”
If it got something right, I built on it: “Yes, that works—now add a spline to the lane and conform it to terrain.”
Each correction became context for the next prompt.
This wasn't just a productivity hack—it’s a new design philosophy.
By treating AI like a collaborator, I was able to:
✅ Generate level design logic
✅ Modularize Blueprint logic
✅ Reduce repetitive in-editor clicking
✅ Prototype faster than ever before
And now, all of that work is wrapped into a clean, accessible free version of the MOBA template.
Ready to start building your own game? My Store has templates, builds, and the prompt list!
🔗 Elcade Studios (@Elcade) | Stan
I’m currently working on a Prompt-Driven Game Design eBook that shows exactly how to use prompts to build projects like this one—from MOBA templates to Tower Defense games and more.
If you're interested in level generation, AI scripting, and context-powered game development, this guide will show you how to build tools for yourself using the same workflows I used here.