How to Save Tokens with a Simple Prompt
One prompt cuts AI verbosity for the entire session — shorter answers, zero filler, same quality output.
Why your AI responses are longer than they need to be
AI tools are trained to be helpful, thorough, and reassuring. That means every response tends to come with an intro, a body, and a summary that repeats the body.
One prompt sent at the start of a session changes this. No settings to find. No configuration. Just paste and go.
The only step — send the low-token prompt
Send the Low-Token Communication skill at the start of any AI session where you want fast, direct output.
Use this skill: Low-Token Communication
What to fill in: Nothing. The prompt works as-is.
When to send it:
- At the very start of a new conversation
- Any time an AI has started giving long-winded responses mid-session (just re-send it)
- Before any task where you need many quick back-and-forth answers
What you get back: “Got it.” — and then for the rest of the session, all responses are short, direct, and free of filler phrases, summaries, and unsolicited explanations.
What it changes
| Without this prompt | With this prompt |
|---|---|
| ”Certainly! That’s a great question. Let me break it down for you…” | Answer. |
| Long response to a yes/no question | ”Yes. [one sentence of context]“ |
| Code + paragraphs explaining what it does | Just the code |
| Summary at the end restating what was just said | Nothing |
What it does not change
- Quality of answers
- Accuracy
- Ability to ask follow-up questions
- Ability to ask for a longer explanation (“explain that in detail” still works)
The prompt turns off the filler. You can always ask for more depth.
Steps: skill primitives in order
Each block is one copy-paste prompt; together they complete the workflow.
Low-Token Mode
Use when AI responses keep adding filler, restating your question, or padding with summaries you did not ask for.
From now on in this conversation, follow these rules exactly: - Answer in as few words as possible - No greetings, no "Great question!", no "Certainly!", no filler phrases - No explaining what you are about to do — just do it - No summarizing what you just said at the end - If I ask for a list, give the list only — no intro sentence, no outro - If I ask a yes/no question, answer yes or no first, then one sentence of context if needed - If I ask for code, give code only — no explanation unless I ask for it - Use plain words. Short sentences. No jargon unless I used it first. - If something is unclear, ask one question only — the most important one. Confirm you understood by replying with just: "Got it."
Related workflows
How to Get Clear Thoughts with AI
Use interview-style prompting to turn a fuzzy idea or problem into something you can actually act on — one question at a time.
Making Your Side Project Live Before It Dies
Most side projects die in planning. This guide uses AI to cut the idea to its core and get it shipped with a concrete step plan.