# Excessive Test Creation - Mostly Harmless? — retained record

> Retained reddit record from reddit:ClaudeAI.

Canonical: https://highsignal.app/data/records/2874fe998cc48631

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeAI)  published 2026-11-15 15:04 ingested 2026-11-16 02:35

[original source ↗](https://www.reddit.com/r/ClaudeAI/comments/1xwxp78/excessive_test_creation_mostly_harmless/)

## retained text



```
One thing I'm sure we've all experienced in coding with Claude is just the sheer volume of tests that get written for each feature.  Test code lines often exceed the actual feature lines.

Over the past 3 months at my company we've switched from primarily hand-written code (with some AI help here and there) to specifically crafted PRDs that drive lengthy coding sessions with Claude code.  Development is primarily on a mid-sized Django app.  Models and migrations are carefully scrutinized to keep the code and logic structure under control, but otherwise quite a lot is done relatively autonomously with a only cursory human review of the code.

Feature productivity has exploded, but a CI suite that used to take 3-4 minutes to run on Github now takes 6-7 minutes.

Some tests seem necessary, but others will often test core Django architecture elements.  I'm curious to get others thoughts - do you spend a lot of time inspecting tests or specifying what you want tested, or do you sort of just ignore the giant test bloat that comes with AI development?
```



document fetched 2026-11-16 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1xwxp78/excessive_test_creation_mostly_harmless
