prompt-master-skill

What this skill does

```markdown
---
name: prompt-master-skill
description: Claude skill that generates accurate, token-efficient prompts for any AI tool — Claude, GPT, Midjourney, Cursor, and 30+ more.
triggers:
  - write me a prompt for
  - generate a prompt for
  - help me prompt
  - fix my prompt
  - create a midjourney prompt
  - write a cursor prompt
  - make a prompt for claude code
  - I need a better prompt for
---

# Prompt Master

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

Prompt Master is a Claude skill that writes sharp, token-efficient prompts for any AI tool. It eliminates the re-prompting loop by extracting full intent on the first pass, routing to the right prompt architecture, and auditing every word for necessity before delivery.

**Supports:** Claude, ChatGPT, Gemini, o1/o3/o4, DeepSeek, MiniMax, Qwen, Cursor, Windsurf, Claude Code, GitHub Copilot, Bolt, v0, Lovable, Devin, Perplexity, Midjourney, DALL-E, Stable Diffusion, ComfyUI, Sora, Runway, ElevenLabs, Zapier, Make, and any unknown tool via universal fingerprint.

---

## Installation

### Recommended — Claude.ai (browser)

1. Download the repo as a ZIP from [github.com/nidhinjs/prompt-master](https://github.com/nidhinjs/prompt-master)
2. Go to **claude.ai → Sidebar → Customize → Skills → Upload a Skill**

### Claude Code (CLI)

```bash
mkdir -p ~/.claude/skills
git clone https://github.com/nidhinjs/prompt-master.git ~/.claude/skills/prompt-master
```

Once installed, the skill activates automatically when Claude detects prompt-generation intent, or you can invoke it explicitly.

---

## How to Invoke

### Natural language (auto-detected)

```
Write me a prompt for Cursor to refactor my auth module
```

```
I need a prompt for Claude Code to build a REST API
```

```
Here's a bad prompt I wrote for GPT-4o, fix it: [paste prompt]
```

```
Generate a Midjourney prompt for a cyberpunk city at night
```

### Explicit invocation

```
/prompt-master

I want to ask Claude Code to build a todo app with React and Supabase
```

---

## What Happens Internally (The Pipeline)

Prompt Master runs a 7-step pipeline on every request. You never see the machinery — only the output.

| Step | What It Does |
|------|-------------|
| 1. **Tool Detection** | Identifies the target AI system and routes to its profile |
| 2. **Intent Extraction** | Pulls 9 dimensions: task, input, output, constraints, context, audience, memory, success criteria, examples |
| 3. **Clarifying Questions** | Asks ≤3 targeted questions if critical info is missing |
| 4. **Framework Routing** | Selects the right prompt architecture from 12 templates |
| 5. **Safe Techniques** | Applies role assignment, few-shot, XML structure, grounding anchors, memory blocks |
| 6. **Token Efficiency Audit** | Strips every word that doesn't change the output |
| 7. **Delivery** | One clean copyable block + one-line strategy note |

---

## Prompt Templates (Auto-Selected)

You never choose a template — Prompt Master picks it. Shown here for understanding:

| Template | Used For |
|----------|----------|
| **RTF** (Role, Task, Format) | Fast one-shot tasks |
| **CO-STAR** | Professional documents, business writing |
| **RISEN** | Complex multi-step projects |
| **CRISPE** | Creative work, brand voice, iterative content |
| **Chain of Thought** | Math, logic, debugging, multi-step analysis |
| **Few-Shot** | Consistent structured output, pattern replication |
| **File-Scope Template** | Cursor, Windsurf, Copilot — code editing AI |
| **ReAct + Stop Conditions** | Claude Code, Devin — autonomous agents |
| **Visual Descriptor** | Midjourney, Stable Diffusion, image AI |
| **Prose Description** | DALL-E 3 |
| **Node Split** | ComfyUI (positive/negative node separation) |
| **Universal Fingerprint** | Any unknown tool — 4 questions, then generates |

---

## Real Usage Examples

### Example 1 — Image prompt (Midjourney)

**You say:**
```
Write me a midjourney prompt for a realistic samurai standing in the rain at night
```

**Prompt Master delivers:**
```
lone samurai standing in heavy rain at night, traditional armor,
neon reflections on wet cobblestone street, cinematic lighting,
dramatic shadows, fog, ultra detailed, photorealistic,
shallow depth of field --ar 16:9 --v 6 --style raw

negative: blurry, low quality, watermark, cartoon, anime, extra limbs
```

> 🎯 Target: Midjourney · ⚡ Framework: Visual Descriptor · 💰 Tokens: Light (~60)
> 💡 Strategy: Comma-separated descriptors over prose, lighting anchored early, version locked, negative prompt prevents style drift.

---

### Example 2 — Coding prompt (Claude Code)

**You say:**
```
Build a claude code prompt for a landing page that looks like Notion
```

**Prompt Master delivers:**
```
Objective:
Build a fully detailed, production-quality static landing page for a
business dashboard product. Design must match Notion's exact aesthetic —
clean, minimal, typographic, with smooth animations throughout.

Stack:
Single HTML file with embedded CSS and vanilla JS. No frameworks.
No external dependencies except Google Fonts (Inter).

Design Spec — match these exactly:
- Background: #ffffff
- Primary text: #1a1a1a
- Secondary text: #6b7280
- Border color: #e5e7eb
- Accent: #000000
- Font: Inter, weights 400 / 500 / 600 / 700
- Base spacing unit: 8px
- Border radius: 6px on cards, 4px on buttons
- Box shadows: 0 1px 3px rgba(0,0,0,0.08) only

Sections to build in order:
1. Navbar — logo left, nav links center, CTA right. Sticky, border-bottom on scroll.
2. Hero — large left-aligned heading, subtext, two CTAs, mockup placeholder.
3. Logos bar — "Trusted by teams at..." with 5 placeholder names, muted gray.
4. Features — 3-column grid, icon + bold title + 2-line description.
   Hover: border darkens to #d1d5db, translateY(-2px), 200ms ease.
5. How it works — alternating left/right, 3 steps.
6. Pricing — 2 tiers. Free: outlined. Pro: black bg, white text, badge.
7. CTA banner — full-width dark section, centered heading, single CTA.
8. Footer — 4-column links, copyright.

Animations:
- Fade-in + translateY(20px→0) on scroll via IntersectionObserver. Threshold 0.15. 500ms ease-out.
- Feature cards: 100ms stagger delay.
- All hover transitions: 200ms ease.
- Smooth scroll on all anchor links.

Constraints:
- Single file only. All CSS and JS embedded.
- No dependencies except Google Fonts CDN.
- Placeholder images = styled gray divs.
- Pixel-perfect at 375px and 1440px.
- Zero console errors.

Done When:
All 8 sections render, animations trigger without layout shift,
hover states work, navbar sticks and border appears on scroll.
```

> 🎯 Target: Claude Code · 💰 Tokens: Medium (~380)
> 💡 Strategy: Every vague aesthetic cue translated into exact hex values and pixel specs — Claude Code cannot guess wrong.

---

### Example 3 — Fix a bad prompt

**You say:**
```
Here's a bad prompt I wrote for GPT-4o, fix it:

"write me some python code to handle user auth"
```

**Prompt Master delivers:**
```
You are a senior Python backend engineer.

Task: Implement a complete user authentication module.

Requirements:
- Framework: FastAPI
- Auth method: JWT (access token 15min, refresh token 7 days)
- Password hashing: bcrypt via passlib
- Endpoints: POST /auth/register, POST /auth/login, POST /auth/refresh, POST /auth/logout
- Store refresh tokens in Redis with TTL
- Return format: { access_token, refresh_token, token_type }

Constraints:
- No third-party auth services (Auth0, Firebase, etc.)
- No ORM — use raw asyncpg queries
- Include input validation with Pydantic v2 models
- Include error handling with proper HTTP status codes

Deliver:
- auth/router.py
- auth/schemas.py
- auth/utils.py
- auth/dependencies.py

Do not scaffold a full app. Auth module only.
```

> 🎯 Target: GPT-4o · ⚡ Framework: RTF + File-Scope
> 💡 Strategy: Vague "some python code" collapsed into exact stack, endpoints, constraints, and file deliverables.

---

### Example 4 — Stable Diffusion

**You say:**
```
Write a stable diffusion