github.devotd.
NewOther

pro-vi/cc-stretch v0.5.2 — A Claude Code plugin that expands in place when you click

The Claude Code plugin pro-vi/cc-stretch, now up to release v0.5.2, shows answers in brief and expands only the phrase you click, right where it is.

Repositorypro-vi/cc-stretch ↗

The Claude Code plugin pro-vi/cc-stretch, now up to release v0.5.2, shows answers in brief and expands only the phrase you click, right where it is.

GitHub description: Stretchtext replies for Claude Code: short by default, each phrase grows in place when you click it

stars
★ 398
Forks
22
Language
TypeScript
License
MIT
Latest release
v0.5.2
Read original source ↗View on GitHub ↗

pro-vi/cc-stretch is a plugin for Claude Code. It is written in TypeScript and released under the MIT license. The repository was created on October 14, 2026, the latest release is v0.5.2, and it has 398 stars. Claude's answers are shown truncated briefly by default. Clicking a truncated phrase expands the following content in place with more length, and pressing the same phrase again collapses it.

This approach is called stretch text (stretchtext). It is a concept laid out by Ted Nelson in 1967: rather than showing everything at once, it extends only the phrase you pressed, in place. According to the author, the reason this idea never spread for a long time is that a person had to write out every step by hand, and having a language model write those steps instead is the starting point of this plugin. The expanded content is passed to Claude as-is, so it does not assume you already read the parts you did not. In a full-screen terminal layout you can operate the expansion with the keyboard and set the display style, and in the Claude desktop app you can click a phrase to expand or collapse it.

The author wrote that the portion visible before expansion was 17% of the whole, and that about two of the three important points were visible without being clicked. This figure was measured against 69 real answers produced in four coding projects, and the measurement method and reproduction procedure were published in a separate directory.

However, the evidence is a measurement the author produced directly in his own projects; no independent evaluation or comparison with other tools is given. What changes v0.5.2 contained also cannot be confirmed. According to the author, most of the code was written by Claude. Because the behavior is tied only to Claude Code's answers, the scope of application is limited in that direction as well. Whether the 17% figure comes out the same in other projects is hard to judge, but the way of expanding and reading only chosen phrases in long output leans toward the useful side.