Prompt engineering, minus the guesswork

Rough prompts in.
Precise specs out.

Paste the prompt you were going to type anyway. Get back a structured spec — objective, context, stack, constraints, output format — aimed at Claude Code, Cursor, Copilot or ChatGPT.

Start free — no cardSee how it works

Two enhancements a day on the free plan · resets at midnight UTC

Input6 words · no structure

add auth to my express api

OutputClaude Code target
Act as a backend / Node.js engineer who ships production code.

# Objective
add auth to my express api

# Context
- Follow the existing project's conventions and structure.
- Do not modify unrelated files.

# Stack
Node.js + Express + TypeScript

# Specifications
- add auth to my express api
- Replace any vague terms with concrete, exact values.

# Constraints
- Keep it focused and single-responsibility; no unused code.
- Handle errors explicitly; validate inputs; no silent failures.

# Output format
Return the complete file(s). Plan briefly first, then implement.
  • Claude CodePlan-first, XML sections
  • CursorConcise, @path aware
  • GitHub CopilotInline-comment intent
  • ChatGPTSelf-contained context
02The problem

A vague prompt doesn't fail loudly.
It just costs you three more.

“Add auth to my express api” gets you something. It compiles. It also picks its own library, invents a user model, skips the refresh flow, and touches four files you didn't ask about.

The fix is never a longer prompt. It's a structured one: role, objective, stack, constraints, output format — the things you would have specified if you'd had a checklist in front of you.

03How it works

Paste, enhance, copy.

  1. 01

    Paste

    Drop in the rough prompt you'd have typed anyway. No forms, no questionnaire.

  2. 02

    Enhance

    Pick your target tool. The prompt is restructured into objective, context, stack, constraints and output format.

  3. 03

    Copy

    Edit anything you don't like, then copy it straight into your editor.

04Capabilities

Small tool, done properly.

It does one job — turning a rough request into a structured prompt — and tries to do that well rather than doing ten things adequately.

01

Built for your actual tool

Claude Code, Cursor, Copilot and ChatGPT each get a different prompt structure — not one generic template with the name swapped.

02

Presets, not questionnaires

One click adds test coverage, error handling, production-readiness or strict typing to the generated spec.

03

See what changed

A diff view highlights everything the enhancer added, so you can judge it rather than trust it.

04

History and saved prompts

Every enhancement is searchable, re-openable and copyable. Keep the ones that worked.

05The obvious question

Why not just ask the model to improve my prompt?

You can, and for a one-off it's fine. The difference is consistency: you burn a round trip, get a different structure every time, and still have to paste the result somewhere. PromptStock applies the same framework every run, targets the specific tool you're using, and keeps the output so you can reuse what worked. It's the difference between asking for a checklist each time and having one.

See the full comparison
06Pricing

Free while you're deciding.

Two enhancements a day on a free account, unlimited on Pro. Cancel and you keep access until the term you paid for ends.

Free

$0 forever

  • 2 enhancements per day
  • Copilot target model
  • Prompt history & saved prompts
Get started

Pro

$9 per month

  • Unlimited enhancements
  • Every target model — Cursor, Copilot, ChatGPT
  • Taxes handled at checkout

Prices in USD. Sales tax / VAT is calculated at checkout. Already subscribed? Manage your plan.

07FAQ

Questions people actually ask.

Full FAQ
Do you train on my prompts?

No. Your prompts are sent to a language-model provider to produce the enhancement and stored against your account so you can search your history. They are not used to train any model, and you can delete them at any time.

Which models does it use?

The enhancement engine sits behind a provider abstraction with a fallback chain, so the underlying model can change without changing your prompts. Bring-your-own-key support is planned, so you can route enhancements through your own provider account.

What are the limits?

A free account gets 2 enhancements a day on the Copilot target. Pro is unlimited across every target model — Claude Code, Cursor, Copilot and ChatGPT. Limits reset at midnight UTC and the workspace always shows how many you have left.

Do I need an account to try it?

Yes. Enhancing requires a free account — it takes an email address and a password, no card. Signing up also means your history and saved prompts are there when you come back.

What happens to my saved prompts if I cancel?

You keep Pro access until the end of the term you already paid for, then move to the free plan. Your history and saved prompts stay — the daily limit changes, your data doesn't.

Is there an API?

Not yet. The enhancement engine is a normal HTTP service, so an API is a natural next step, but it isn't available today and isn't part of any paid plan.

Try it on the next prompt you were about to write.

Free account · no card · close the tab if it doesn't beat what you typed

Start free