# I built a Claude Code mod on Haiku 5.5 (or Jev) that picks the effort per prompt (and much more) — retained record

> Retained reddit record from reddit:ClaudeCode.

Canonical: https://highsignal.app/data/records/f4b98db1d3b894a1

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeCode)  published 2026-10-09 02:06 ingested 2026-10-10 02:37

[original source ↗](https://www.reddit.com/r/ClaudeCode/comments/1x19b2c/i_built_a_claude_code_mod_on_haiku_55_or_jev_that/)

## retained text



```
**What it is:** effortless, a mod that puts a bar above the prompt. Haiku 5.5, or Jev if you have a TypeSafe key, reads each prompt and sets the effort. On prompts it judges simple, it can run that one prompt on Haiku 5.5 instead of your chat’s model. Your chat and its cache stay where they were. The bar also warns when your prompt cache goes cold and offers a handoff when you’ve moved on to something new.

**How Claude Code was used:** it was built in Claude Code sessions, with me steering and testing each build in the desktop app. Claude also cut the launch video and wrote the copy.

**What I learned:**

* In about 80k of my own requests, cache reads were around 76% of the cost and output only 8%. Effort mostly changes how many tool calls a prompt makes, not how long the answers are.
* Switching effort keeps the cache on Opus 5.5 and Sonnet 5.5. On Fable 5.1 and older Opus it rewrote 56 to 100% of it, so Auto pauses there.
* Short replies like “yes” after a question fooled the judge at first. Now they always go to it.

Free, MIT, not affiliated with Anthropic. Tested in the desktop app.

[https://heycubit.github.io/effortless/](https://heycubit.github.io/effortless/)
```



document fetched 2026-10-10 00:40 · canonical https://reddit.com/r/ClaudeCode/comments/1x19b2c/i_built_a_claude_code_mod_on_haiku_55_or_jev_that
