ChatGPTGeminiClaudeGrok
Paste a rough prompt and get it back tidied and placed inside a labeled scaffold: Role, Task, Format; Context, Task, Constraints, Format; or a step-by-step reasoning block. Your words are never rewritten, and nothing leaves your browser.
Formatter free forever | Prompt library with 2 saved prompts free inside ChatGPT, Gemini, Claude and Grok | Premium $9.99/mo or $99 once
Framework
Adds a Role heading and an Output format heading around your text. Best for most everyday prompts.
Your words are kept exactly; the tool only normalizes whitespace and adds labeled sections with {{placeholders}}. Nothing you paste is uploaded. Token estimate assumes about 0.75 words per token; for an exact count use the token counter.
Why structure
A model treats each heading as a separate instruction. The same request written as one paragraph makes it guess which sentence is the task, which is the constraint and which is the format.
The scaffolds here are the ones that hold up across providers: a role, a task, a format, and for harder work a context block and a constraints list. None of them depend on a specific model, which is why a prompt formatted once can be pasted into ChatGPT, Claude, Gemini or Grok unchanged.
The tool deliberately does not write the role or the format for you. Those are the parts that carry your intent, and a guessed role is worse than an empty field you fill in. The counter on the right shows how many are still blank.
Four principles
What the scaffold is doing:
Fill every field before you send: an unfilled {{role}} is read literally by the model.
Step by step
Four steps, no account, and nothing leaves your browser.
Drop in the rough version, notes, half sentences and stray blank lines included. The text stays in your browser; nothing is sent anywhere.
Role, Task, Format covers most prompts. Choose Context, Task, Constraints, Format for high-stakes work, the reasoning scaffold for analysis, or Clean only to tidy without a scaffold.
Your words land under Task, untouched. Every {{field}} in the scaffold is a blank the tool cannot fill for you: the role, the format, the constraints. The counter shows how many are left.
Copy the result into ChatGPT, Claude, Gemini or Grok. To reuse it, save it in the AI Toolbox prompt library, where the same {{fields}} become dynamic values you fill at insert time.
The frameworks
Four options, from a full context-and-constraints layout to a plain whitespace tidy. The chips in the tool switch between them.
| Framework | Best for | Structure | What the tool adds |
|---|---|---|---|
| Role, Task, Format Selected | Most everyday prompts | Who the model is, what to do, how to answer | Adds a Role heading and an Output format heading around your text |
| Context, Task, Constraints, Format | Complex or high-stakes tasks | Adds background and explicit limits before the format | Adds Context, Constraints and Output format headings around your text |
| Task + step-by-step | Analysis, math, debugging | Asks the model to show its work before the answer | Adds an Instructions heading that asks for reasoning first |
| Clean only | Prompts you have already structured | Trims whitespace, no scaffold | Only normalizes whitespace and blank lines |
Want a starting prompt rather than a scaffold? Browse the prompt templates.
In the chat itself
This page formats one prompt. AI Toolbox keeps the result inside ChatGPT, Gemini, Claude and Grok, ready to insert.
Reuse
A formatted prompt is worth keeping. AI Toolbox holds one prompt library across ChatGPT, Gemini, Claude and Grok: save the result, type two slashes in any compose box to insert it, and keep the {{fields}} as dynamic values that a small form collects each time. The free plan holds two saved prompts and five from the catalog.
Plan: Free: 2 saved prompts, 5 catalog prompts. Premium: unlimited.
Sequence
Long jobs are a sequence of prompts: outline, draft, edit, summarize. Premium chains run them in order from one trigger, two dots in the compose box, with a shared {{topic}} variable filled once at the start. Each step is a prompt you formatted here.
Plan: Premium: prompt chains with shared variables on all four platforms.
Use cases
Keep what you format
Install AI Toolbox for a prompt library you insert with two slashes on ChatGPT, Gemini, Claude and Grok, with dynamic fields filled at insert time.
Add to Chrome, freePricing
Two saved prompts, five catalog prompts, folders and search are on the free plan. Premium adds unlimited prompts, prompt chains, full-text search and export. Not an AI subscription: you keep using your own accounts.
Free
$0forever
Premium
$9.99/month
LifetimeMost popular
$99one-time
Pays back in under 10 months of Premium.
Get Lifetime AccessAll Access LifetimeBest value
$199one-time
Teams$15/seat/month
All Premium features across the four modules · Shared team prompts · Admin dashboard · $12/seat/mo billed annually · 14-day free trial
Refunds follow the refund policy; renewals are not refundable. Premium unlocks for the email you enter at checkout, so use the same address you sign in to ChatGPT with.
FAQ
It normalizes whitespace (trailing spaces, Windows line endings, runs of blank lines) and, if you choose a framework, places your text under a Task heading with labeled sections around it: Role and Output format, or Context, Constraints and Output format, or an Instructions block that asks for step-by-step reasoning. It runs entirely in your browser with no AI call, so nothing you paste leaves the page.
No. It never rewrites, shortens or reorders your words. The only edits are whitespace normalization and, with Bullet each line on, a leading dash on each non-empty line. Everything else in the output is scaffold text and {{placeholders}} you fill in yourself.
A three-part structure: who the model should be (Role), what it should do (Task) and how it should answer (Format, including length, tone and layout). An explicit role and an explicit output format are the two instructions models follow most consistently, which is why RTF is the default here.
When the model needs background it cannot infer, or when there are rules it must not break: a word limit, a banned claim, a required section, a house style. Listing constraints as their own bullets is more reliable than folding them into the task sentence.
An Instructions block asking the model to think through the task before answering and to put the final answer under an Answer heading. It helps on analysis, math and debugging, where the model otherwise jumps to a conclusion; it adds length on simple factual prompts, so leave it off there.
Yes. The frameworks are plain text with Markdown headings, which all four read the same way. If you want the same prompt available in all four, AI Toolbox keeps one shared prompt library you insert by typing two slashes in the compose box on any of them.
Copy it into the AI Toolbox prompt library. Saved prompts keep their {{fields}} as dynamic values that a small form collects at insert time, so the wording stays identical run to run. The free plan holds 2 saved prompts and 5 from the catalog; Premium ($9.99/month or $99 once per module) makes them unlimited and adds prompt chains.
Free, with no account or sign-up, and nothing is uploaded: the formatting is a few string operations in your browser. The AI Toolbox extension is also free to install; the prompt library with two saved prompts, folders and search are on the free plan.
40,000+ users · 4.7/5 on the Chrome Web Store
Install AI Toolbox for a prompt library on ChatGPT, Gemini, Claude and Grok, inserted with two slashes, with fields you fill at insert time.
ChatGPT
Gemini
Claude
Grok