<div align="center"> <picture> <source media="(prefers-color-scheme: dark)" srcset="assets/logo-dark.svg" /> <img src="assets/logo.svg" alt="hush" width="240" /> </picture> <h1>hush</h1> <p><strong>Shuts Claude up so your session stops costing money to read.</strong></p>

<img src="assets/bench-narration.svg" alt="Every session in the benchmark suite drawn as a waveform, one spike per run, amplitude is words of play-by-play before the answer. The no-plugin lane spikes across half the suite and peaks at 76 words, silent in 17 of 32 sessions. The hush lane is close to a flat line — silent in 27 of 32, with the remaining spikes on the bug-hunting tasks" width="700" />

<p><em>This is what a session sounds like.</em></p> </div>

[](./LICENSE) [](https://docs.anthropic.com/en/docs/claude-code)

---

## What is this?

You've seen it: "Let me start by looking at the codebase." "Now I'll check the config." Four hundred lines of build output you didn't ask for, followed — eventually — by the one sentence you actually needed. Every word of that is billed. All of it.

hush doesn't ask Claude to "be more concise" and hope for the best. It trims the actual bulk — logs, command output, narration — at the source, as it happens, before any of it hits your bill.

It's built for real engineering sessions — the kind that read logs, run builds, and dig through output — because that's where the noise actually lives.

## Why you'd want it

- **Cheaper sessions.** It shrinks the two biggest sources of bulk — noisy output and narration — so long sessions cost less. - **Easier to read.** The answer sits at the top of one final message, not buried in a play-by-play. - **Nothing important is lost.** Failing command output, code, diffs, and security warnings are kept whole. - **Zero setup.** Install it and it's on. Tune it later only if you feel