freemcplab / Verzeichnis / Playground / prompt-master

● Community nidhinjs ⚡ Sofort

prompt-master

Name: prompt-master MCP Server
Author: nidhinjs

A meta-skill that turns vague requests into tight, effective prompts for Claude, ChatGPT, Midjourney, Cursor, and 30+ other AI tools.

Prompt Master runs a 6-step pipeline on your ask: detect target tool, extract intent across 9 dimensions, clarify if needed, pick from 12 prompt templates, audit for waste, and deliver a clean copy-ready prompt. It preserves prior decisions in a Memory Block so your follow-ups don't contradict the setup.

▶ Demo ansehen 📦 Installieren 💡 Anwendungsfälle

Warum nutzen

Hauptfunktionen

Supports 30+ target tools (Claude, ChatGPT, Midjourney, DALL-E, Cursor, Copilot, etc.)
Detects 35 'credit-killing' patterns — vague verbs, missing success criteria, assumed context
Memory Block preserves prior decisions across the session to avoid contradictions
Applies only proven techniques (role assignment, few-shot, XML structure); excludes unreliable ones
Asks up to 3 clarifying questions rather than guessing

Live-Demo

In der Praxis

prompt-master-skill.replay ▶ bereit

0/0

Installieren

Wählen Sie Ihren Client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "prompt-master-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/nidhinjs/prompt-master",
        "~/.claude/skills/prompt-master"
      ],
      "_inferred": true
    }
  }
}

Öffne Claude Desktop → Settings → Developer → Edit Config. Nach dem Speichern neu starten.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "prompt-master-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/nidhinjs/prompt-master",
        "~/.claude/skills/prompt-master"
      ],
      "_inferred": true
    }
  }
}

Cursor nutzt das gleiche mcpServers-Schema wie Claude Desktop. Projektkonfiguration schlägt die globale.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "prompt-master-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/nidhinjs/prompt-master",
        "~/.claude/skills/prompt-master"
      ],
      "_inferred": true
    }
  }
}

Klicken Sie auf das MCP-Servers-Symbol in der Cline-Seitenleiste, dann "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "prompt-master-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/nidhinjs/prompt-master",
        "~/.claude/skills/prompt-master"
      ],
      "_inferred": true
    }
  }
}

Gleiche Struktur wie Claude Desktop. Windsurf neu starten zum Übernehmen.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "prompt-master-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/nidhinjs/prompt-master",
        "~/.claude/skills/prompt-master"
      ]
    }
  ]
}

Continue nutzt ein Array von Serverobjekten statt einer Map.

~/.config/zed/settings.json

{
  "context_servers": {
    "prompt-master-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/nidhinjs/prompt-master",
          "~/.claude/skills/prompt-master"
        ]
      }
    }
  }
}

In context_servers hinzufügen. Zed lädt beim Speichern neu.

claude mcp add prompt-master-skill -- git clone https://github.com/nidhinjs/prompt-master ~/.claude/skills/prompt-master

Einzeiler. Prüfen mit claude mcp list. Entfernen mit claude mcp remove.

Anwendungsfälle

Praxisnahe Nutzung: prompt-master

Craft a Midjourney prompt that actually matches your mental image

👤 Designers and marketers working with image models ⏱ ~5 min beginner

Wann einsetzen: Your first 3 Midjourney generations came back wrong and you're burning fast-hours.

Ablauf

Describe what you want in plain language

Use prompt-master. I want a hero image for a fintech landing page — serious, warm, not boring.✓ Kopiert

→ Claude asks 1-2 clarifying questions about aspect ratio and style anchors
Get the optimized prompt

Yes — 16:9, inspired by Monocle magazine.✓ Kopiert

→ A Midjourney-ready prompt with weights, --ar, --style, --s params

Ergebnis: A reusable prompt you can vary slightly for a full campaign.

Fallstricke

Skill over-clarifies for a simple request — Say 'skip clarification, one-shot it'

Write a tight Cursor system prompt for a recurring refactor

👤 Developers who use Cursor for codebase-wide changes ⏱ ~10 min intermediate

Wann einsetzen: You're about to run the same refactor across 20 files and want a prompt that actually constrains the model.

Ablauf

State the refactor

prompt-master — write a Cursor prompt: convert all React class components in src/legacy to functional components with hooks, preserve prop types, no behavioral changes.✓ Kopiert

→ Structured prompt with explicit constraints, negative examples, output format
Test on one file first

Apply the prompt to src/legacy/UserList.tsx and show the diff. Does it hold?✓ Kopiert

→ Clean diff that matches the stated constraints

Ergebnis: A prompt you can paste into Cursor and trust across the full directory.

Fallstricke

Prompt too generic, model freelances — Ask prompt-master to add at least one concrete before/after example

Kombinationen

Mit anderen MCPs für 10-fache Wirkung

prompt-master-skill + nano-banana-pro-prompts-recommend-skill

Use prompt-master to structure the intent, then the Nano Banana skill to pick a tested image prompt pattern

Use prompt-master to nail down what I want, then hand off to nano-banana-pro-prompts for the final image prompt.✓ Kopiert

Werkzeuge

Was dieses MCP bereitstellt

Werkzeug	Eingaben	Wann aufrufen
Tool detection	user request	First stage of the pipeline
Intent extraction	request + context	After tool detection
Clarification	missing-critical-info flags	Only when success criteria unclear
Framework selection	task type	Before composition
Efficiency audit	draft prompt	Final polish step

Kosten & Limits

Was der Betrieb kostet

API-Kontingent: None
Tokens pro Aufruf: 1-3k per prompt optimization — small
Kosten in €: Free — skill is local prompts
Tipp: The whole point of the skill is reducing wasted tokens downstream; the upfront cost is tiny.

Sicherheit

Rechte, Secrets, Reichweite

Credential-Speicherung: No credentials — skill is pure prompts

Datenabfluss: None

Fehlerbehebung

Häufige Fehler und Lösungen

Skill over-asks clarifying questions

Add 'make reasonable assumptions, don't ask more than 1 question' to your request.

Output prompt is too long

Explicitly request 'efficient mode' — the audit step will strip harder.

Memory Block loses context between sessions

Memory is per-working-directory; run follow-ups from the same folder.

Alternativen

prompt-master vs. andere

Alternative	Wann stattdessen	Kompromiss
Manual prompting with a template library	You already know the patterns and just need a reference	Faster for experts; prompt-master wins for non-obvious cases

Mehr

Ressourcen

📖 Offizielle README auf GitHub lesen

🐙 Offene Issues ansehen

🔍 Alle 400+ MCP-Server und Skills durchsuchen