# Codex Mortis began as a test of giving AI tools access to game development

> Before it was a game, the project was a question about which parts of making a game an agent could operate.

Published 2026-11-08

Canonical: https://machines-on-machines.anth.us/posts/codex-mortis-development-pipeline/

A player asked how Codex Mortis took shape. In a pinned Steam discussion, a developer described it as an experiment that began after trying AI tools in existing projects. The agents could reach some parts of development but not much of the engine. The team wanted to know what would happen if it built a workflow with agent access in mind from the start.

::image{
  src="assets/codex-mortis-development-pipeline/mortis-6.png"
  layout="inline"
  alt="An isometric Codex Mortis battle with a red attack zone across the floor."
  caption="A battle in the release build of Codex Mortis, the game that grew out of the experiment."
  credit="GROLAF / CRUNCHFEST. Publisher-supplied press screenshot."
}

[Image source](https://drive.google.com/file/d/1lDNbolG3lQqDoovVZPO1ut0KMQ0gmN1_/view) · [Reuse terms](https://drive.google.com/drive/folders/1il2Ml_SeY4t9JbBqt8PdH4zAAEID_n2o)

The distinction is about what a tool can do after making a suggestion. An agent can write ideas in a chat window without access to the files or interfaces that turn them into game mechanics. Building a workflow around it means deciding what those interfaces expose and how the team can check changes.

The account comes from a developer responding to players. It explains the starting point for one project and gives other creators a concrete question to ask about AI tools. Which parts of the game can the agent actually change?

[Read the source](https://steamcommunity.com/app/4084120/discussions/0/842866071372380925/).
<!-- papyrus-source-version: 3 -->
