# Backwards overthunk compatibility

**URL:** <https://community.openai.com/t/backwards-overthunk-compatibility/1405415>\
**Category:** Codex\
**Tags:** codex\
**Created:** [October 11, 2026, 7:21pm UTC](https://community.openai.com/t/backwards-overthunk-compatibility/1405415 "2026-10-11T19:21:33Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![anvilz](https://avatars.discourse-cdn.com/v4/letter/a/f19dbf/32.png) [@anvilz](https://community.openai.com/u/anvilz)\
**Post date:** [October 11, 2026, 7:21pm UTC](https://community.openai.com/t/backwards-overthunk-compatibility/1405415/1 "2026-10-11T19:21:33Z")

</div>

I start a brand-new repo, and within an hour, Codex is already trying to preserve earlier code and design decisions for backward compatibility. Five hours in, “legacy” compatibility code has crept into a dozen places.

Is there something in Codex’s system prompts that makes it treat every repo as a production codebase, even when it’s still an early prototype? Why does it assume backward compatibility is already a requirement?