Built for Claude · macOS · Windows · Linux

Better prompts.
One hotkey.

Select your draft, press ⌃⌥⌘P (or Ctrl+Alt+P on Win/Linux), and watch it turn into a structured Claude prompt — with role, context, and output format — right in the field where you're typing.

Download for macOS

Free forever · No account required · macOS 13+ · Windows 10+ · Linux (AppImage/.deb)

claude.ai/new
Your draft
help me with python parser for nested json
After hotkey → 
<role>experienced Python developer</role>
<task>parse nested JSON structure</task>
<instructions>
  - use json + jsonpath-ng
  - handle missing keys
  - example output
</instructions>
<output_format>working code + 3 examples</output_format>

Demo video coming soon

Works in

Claude.aiChatGPTNotionSlackFigmaVS CodeTelegramLinearSuperhumanArcSafariObsidian

You write a draft.
Claude writes you a question back.

Most prompts fail at the first step: they describe what you want, not what the model needs to produce a useful answer. So you iterate. Rephrase. Try again. Lose twelve minutes.

You typed
"help me with python"
You got back
a clarifying question and three bullet points of context you already have

The gap between what you mean and what the model understands is not a model problem. It is a prompt structure problem.

Three steps. No context switch.

01

Select your draft

Highlight any text in any app. A rough sentence, a half-formed question, a wall of stream-of-consciousness — whatever you have.

02

Press your hotkey

Hit ⌃⌥⌘P (fully customizable). Prompt Enhancer captures your text, sends it to Claude Haiku via your API key, and applies the Anthropic XML structure.

03

Watch it transform

Your draft is replaced in place — in the same field, in the same app — with a structured prompt ready to send. Under four seconds.

Built for every kind of prompt

For developers

In Cursor, Claude Code, VSCode

Draft

fix the auth bug

Enhanced
<role>senior backend dev</role>
<task>find auth middleware bug</task>
<context>JWT validation fails for refresh tokens</context>
<instructions>
  - check token expiry logic
  - test edge cases
</instructions>
For marketers

In Claude.ai, ChatGPT, Notion

Draft

write me email about launch

Enhanced
<role>experienced marketer</role>
<task>write launch announcement email</task>
<context>SaaS product, 5000 beta users, paid plan goes live</context>
<output_format>
  - subject line (max 50 chars)
  - body 150-200 words
  - clear CTA
</output_format>
For designers

For Midjourney, DALL-E, image AI

Draft

cool hero image for landing

Enhanced
<role>art director</role>
<task>create hero image prompt</task>
<style>minimal, dark, brutalist typography over abstract gradient</style>
<technical>16:9 aspect, --ar 16:9 --v 6 --style raw</technical>
For PMs

In Linear, Notion, Slack

Draft

write user story for new feature

Enhanced
<role>senior product manager</role>
<task>write user story for feature X</task>
<format>
  - As a [user], I want [action], so that [benefit]
  - acceptance criteria
  - edge cases
</format>

Pay once. Use forever.

No subscriptions. No accounts. License keys bound to your platform — macOS, Windows, or Linux.

Free

Try it without setup

$0 forever
  • ✓ 20 free demos with our key
  • ✓ Unlimited with your own Claude key (BYOK)
  • ✓ All 4 presets (Default, Concise, Verbose, Code)
  • ✓ macOS · Windows · Linux builds
Download for macOS
Early bird — first 100

Lifetime Pro

Everything, forever — license key issued for your OS

$19 $29 once
  • ✓ Unlimited managed (we pay Anthropic)
  • ✓ Platform-specific license key (PE-MAC-… / PE-WIN-… / PE-LNX-…)
  • ✓ Activate on up to 3 devices on chosen OS
  • ✓ All presets + premium domain-specific presets
  • ✓ Full enhancement history · Priority support
Buy Lifetime — macOS

Need multi-OS bundle? Email us — $29 covers all three platforms.

Compare: BoltAI $69 · Kerlig $79 · Raycast $192/year · Prompt Enhancer $19-29 once

Windows · Linux waitlist

macOS ships now. Windows and Linux builds drop Q3 2026 — get notified.

Your prompts stay private

BYOK mode: direct to Anthropic. Pro Managed mode: through our zero-log proxy. You choose.

API key in OS Keychain

Same encryption your browser uses for passwords. Never leaves your device.

BYOK mode: direct to Anthropic

Your prompts go directly to Anthropic — we never see them. BYOK is the default and works on Free and Pro.

Pro Managed mode: zero-log proxy

Prompts route through our zero-log proxy (TLS only, no content storage). We log only: timestamp, token count, license ID — never the text.

No prompt content logged

Local logs record only metrics: length, tokens, latency. Not your text. Not on our servers.

Questions

Which platforms are supported?
macOS 13+ (Apple Silicon + Intel), Windows 10/11 (x64 + ARM64), Linux x86_64 (AppImage + .deb for Ubuntu/Debian/Fedora). The website auto-detects your OS and offers the right download. License keys are platform-specific (PE-MAC-…, PE-WIN-…, PE-LNX-…) and bind to the OS you bought for.
Is it safe to give my Claude API key?
Your key is stored in the OS secure credential store — macOS Keychain, Windows Credential Manager, or libsecret / GNOME Keyring on Linux. It never leaves your device except to talk to api.anthropic.com. Nothing is sent to our servers.
What happens to my prompts?
In BYOK mode (default): your prompt goes directly to Anthropic — we never see it, never store it. In Pro Managed mode: your prompt routes through our zero-log proxy (TLS only). We log only timestamp, token count, and your license ID — never the prompt content.
What does each preset do?
Default — universal enhance: adds role, constraints, output format. Concise — strips extras, optimises for minimal-token output (good for ChatGPT-mini and Claude Haiku). Verbose — expands reasoning steps, adds counter-arguments and edge-cases (good for Opus / o1-style models). Code — for programming prompts: detects language/framework, asks for output expectations, test cases.
Does it work in [my app]?
Any app with a text field. macOS uses Accessibility API, Windows uses UI Automation, Linux uses AT-SPI; Electron / web apps on all 3 platforms fall back to clipboard automatically.
What if my OS updates break it?
macOS app is native Swift (macOS 13+); Windows/Linux app is built on Tauri (Rust). Compatibility patches ship within 2 weeks of major OS releases.
Can I use one license on macOS AND Windows?
No — each $19 license is bound to one platform (your OS family at purchase). Multi-platform bundle is $29 — email support@promptenhancer.online and we issue all 3 keys. Each key activates on up to 3 devices of its OS family.
Money back?
30-day refund, no questions asked. Email support@promptenhancer.online.
Works offline?
No — enhancement requires Claude API. You'll see a clear "no internet" message if disconnected.
How is this different from Kerlig / Raycast / BoltAI?
Prompt Enhancer does one thing and does it well: structures prompts using the official Anthropic XML format. Kerlig is a general AI writing assistant. Raycast is a launcher with AI add-ons. BoltAI is a multi-model client. We are specialized.
Why is it free? What's the catch?
You bring your own Claude API key — you pay Anthropic directly, not us. Free tier (20 demo calls) uses our key. Pro Lifetime $19-29 gives you unlimited managed access + premium features on your chosen OS.

Make Claude
actually understand you.

Available on macOS, Windows and Linux. One hotkey. Same prompt engine.

Download for macOS

macOS 13+ · Windows 10+ · Linux x86_64 · No account · Free forever (BYOK)