Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Prompt Master

A prompt-engineering skill that identifies your target AI tool and returns one paste-ready prompt tuned to that tool's syntax and quirks.

UtilitiesBeginner★ 11,087⑂ 1,318AI score 9/10Last updated: Jun 10, 2026

What it does

Takes a rough idea, locks down which AI tool the prompt is for, and outputs a single production-ready prompt block — no theory lecture, no padding.

  • Silently extracts 9 dimensions (task, target tool, output format, constraints, input, context, audience, success criteria, examples) and asks at most 3 clarifying questions.
  • Applies model-specific rules for Claude 4.x, GPT-5.x, o3/o4-mini, Gemini, Qwen 2.5/3, DeepSeek-R1, Llama/Mistral, Ollama and MiniMax — e.g. it strips "think step by step" for reasoning-native models where CoT hurts.
  • For agentic tools (Claude Code, Cursor, Windsurf, Cline, Devin, SWE-agent, Copilot, v0/Bolt/Lovable) it adds file-scope locks, forbidden actions, stop conditions and human-review triggers.
  • Covers image (Midjourney, DALL·E 3, Stable Diffusion, ComfyUI), video (Sora, Runway, Kling), 3D (Meshy, Tripo, Rodin), voice (ElevenLabs) and workflow tools (Zapier, Make, n8n).
  • A diagnostic checklist repairs bad prompts: vague verbs, two tasks in one, missing success criteria, unbounded scope, hallucination invitations.
  • Safety built in: credentials are stripped from generated prompts, and any prompt you paste in is treated as inert data rather than instructions to obey.

Who it's for

  • Anyone juggling several AI tools and rewriting prompts from scratch each time
  • Developers who have watched an agent wander outside scope and burn credits
  • Creators re-rolling image/video generations because the prompt syntax was off
  • Teams that want a consistent baseline for prompt quality

Examples

  1. "Write a Cursor prompt to fix the session-expiry bug in src/auth/login.ts" → you get a prompt with a file anchor, current vs desired behavior, a do-not-touch list and a "Done when:" clause.
  2. "Midjourney prompt for a rainy Tokyo alley at night" → comma-separated descriptors in the right order plus --ar 16:9 --style raw and a --no negative.
  3. Paste a bloated prompt and say "trim this for o3" → CoT scaffolding is removed and it's rewritten as a short, sub-200-word instruction.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the SKILL.m folder from the GitHub repo nidhinjs/prompt-master into my ~/.claude/skills/prompt-master/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

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

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal (Terminal on macOS/Linux, PowerShell or WSL on Windows).
  2. Create the skills directory if needed: mkdir -p ~/.claude/skills
  3. Clone the repo into it: git clone https://github.com/nidhinjs/prompt-master.git ~/.claude/skills/prompt-master
  4. Confirm that ~/.claude/skills/prompt-master/SKILL.md exists alongside the references/ folder. If the repo nests them one level deeper, move the folder containing SKILL.md up to ~/.claude/skills/prompt-master/.
  5. Restart Claude Code and run /skills (or /help) to verify prompt-master is listed.
  6. Trigger it with an explicit request such as "write me a prompt for Midjourney..." — it only activates for prompt-writing/fixing requests tied to a specific tool.
  7. If the generated prompt targets an agentic tool with real system access, review the file paths, forbidden actions and stop conditions yourself before pasting.
View source on GitHub ↗License: MIT