← sources

I built my own engine for a one-hour horror game. Here’s why it actually made sense

source: reddit (reddit:gamedev)published 2026-10-03 16:22ingested 2026-10-04 02:36

retained text

Building a custom engine for a one-hour indie horror game sounds like a terrible scope decision.

And honestly, at first it probably was.

But the architecture ended up solving a problem I kept running into: I didn’t want every scare, door, conversation, puzzle, and scripted event hardcoded into the level.

So I separated the game into three pieces:

**Stage:** the 3D environment  
**Script:** a declarative story.json  
**Engine:** a runtime that interprets the story

The engine now understands 77 different types of actions: dialogue, sounds, doors, objectives, animations, enemy behavior, camera effects, state changes, waits, conditions, and so on.

The actual game contains 70 sequences, 71 triggers, 116 interactables, and 21 documents.

One of the biggest unexpected benefits was tooling. Because the story is structured data, I can validate references before running the game, detect missing sounds/doors/voice lines, validate navigation, and simulate game state to jump directly into later chapters.

It also created a useful separation for horror design:

The engine controls *how* something is frightening.  
The story controls *why and when* it happens.

I wrote up the architecture and some examples here:

[https://0xkot.itch.io/the-cradle-episode-01/devlog/1687752/why-i-built-my-own-engine-for-a-one-hour-horror-game](https://0xkot.itch.io/the-cradle-episode-01/devlog/1687752/why-i-built-my-own-engine-for-a-one-hour-horror-game)

I’m curious if anyone else working on narrative-heavy games has gone down a similar data-driven route instead of scripting events directly inside scenes.

document fetched 2026-10-04 00:24 · canonical https://reddit.com/r/gamedev/comments/1wwraod/i_built_my_own_engine_for_a_onehour_horror_game