"/no-fluff" Claude Skill
One command that makes Claude say the last thing again, shorter and in a way that makes sense with plain words.
You ask a good question and you get back nine paragraphs, three headings and a table. Somewhere in there is the answer. You read it twice, get the gist, and move on without ever being sure you understood the middle part.
That is not a knowledge problem. The information was all there. It arrived in a shape you could not absorb at the speed you needed it.
This is one command that fixes it. Type /no-fluff and Claude says the same thing again, smaller and plainer.
What it is not
It is not a summary. A summary drops things, and the thing it drops is usually the thing you needed. This keeps every fact that changes what you decide, and throws away the packaging around it: the preamble, the context you already had, the options it already ruled out, the caveats that do not change the answer.
If something did get dropped, it tells you. One line at the bottom saying what was cut, so you know what to ask about next instead of assuming you have the whole picture.
The part that "be more concise" gets wrong
Telling an AI to be brief usually gets you the same jargon in fewer words. Just as impenetrable, now with less to go on.
This file is explicit that the job is to rewrite the language, not only the length. Replace the term or define it in three words. Lead with the answer instead of building up to it. Keep every number, name and file path exactly as it was, because those are the parts most worth keeping.
It changes the rest of the conversation
Run it once and Claude stays in that register for the rest of the session. You are telling it how you want to be spoken to, not just asking for one message back.
What it costs
Nothing. It is a text file. It works in Claude Code and in the Claude app, on any project, in any conversation.
Where it works
Anywhere you talk to Claude. It is one file with nothing else attached, so it behaves the same in Claude Code and in the Claude app.
Option A: Claude Code
If you have not got Claude Code yet:
- 1Install Node.js from nodejs.org, if
node --versionsays nothing. - 2Run
npm install -g @anthropic-ai/claude-codein a terminal.
Then unzip the download into your skills folder:
- Mac or Linux:
~/.claude/skills/ - Windows:
C:\Users\<you>\.claude\skills\
You should end up with .claude/skills/no-fluff/SKILL.md. If the .claude folder is not there yet, make it.
Putting it in that folder rather than inside one project is deliberate. It makes the command available in every project you ever open, which is the whole point of this one.
Restart Claude Code so it picks the skill up, then type /no-fluff any time a reply is more than you wanted to read.
Option B: the Claude app
Go to Settings, then Capabilities, then Skills, and upload the zip. Start a new chat. When an answer is too long, ask for the no fluff version and it will pick the skill up.
Using it
Run it straight after the message you could not get through. It works on the last thing Claude said.
You can also point it at one part: /no-fluff just the pricing bit. That narrows it to the section you are actually stuck on instead of restating everything.
One thing worth knowing
If you find yourself running it over and over, that is a signal rather than a fix. The file ends by telling Claude to stay in that register for the rest of the session, so running it once early in a long conversation does more than running it five times late in one.