wtf is justsaydone?

Justsaydone is a plugin for Claude Code, Anthropic's coding assistant, that changes how it talks after finishing a task, not how it does the work. Normally, after fixing a bug or creating a file, Claude Code writes a paragraph explaining what it changed and offering to do more. This plugin turns that paragraph into a single word, done, once you have installed it and started a fresh session. The idea behind it is that most of that closing summary just repeats what your diff, your test output, and your own request already told you, and reading it costs time and output tokens. Claude still plans, edits, and verifies its work exactly as before. It just stops narrating the result once the task is finished and confirmed successful. There is a built in escape hatch: if your next message contains words like explain, summarize, why, what did you change, or details, Claude gives a full normal answer for that one turn, then goes back to replying with done afterward. Questions still get complete answers rather than being cut short. The project includes a benchmark suite that runs the same tasks with and without the plugin and measures reply length and total token usage. On tasks where Claude does work and reports back, replies shrink by roughly 90 to 95 percent in word count, though total session tokens change less since most of a coding turn's output is the code itself, which this plugin does not touch. One rule overrides brevity: Claude only says done when the work actually succeeded and was verified. If a task fails or gets blocked, it explains what went wrong in one line instead of falsely claiming success. Under the hood, justsaydone is a single Claude Code output style packaged as a plugin that turns itself on automatically once installed, while keeping every one of Claude Code's normal engineering behaviors intact. Installation is done through Claude Code's plugin marketplace commands, and the mode can be turned off again through the plugin menu. It is licensed under MIT.