Prompt Master

prompt-master (v1.7.0) is a specialized tool designed to generate optimized prompts for various AI tools, including LLMs, Cursor, Midjourney, image/video AI, and coding agents.

Activation Logic: It only activates when you explicitly ask to write, fix, improve, or adapt a prompt. It does not activate for general conversation, standard coding tasks, or document writing.

When prompt-master generates a prompt, it follows a strict output structure to ensure it is production-ready and easy to use:

1. A single copyable prompt block: The optimized prompt ready to be pasted into the target tool.
2. Target and Optimization Summary:
   • 🎯 Target: [tool name]
   • 💡 [One sentence explaining what was optimized and why]
3. Setup Instructions (Optional): If the prompt requires specific setup steps, a short 1-2 line plain-English note will be provided below the summary. Only included when genuinely necessary.

For copywriting or content-related prompts, the output will include fillable placeholders such as [TONE], [AUDIENCE], [BRAND VOICE], and [PRODUCT NAME] where appropriate.

Prevent the AI from overreaching or touching irrelevant files by defining strict boundaries.

Key improvements include:

• Set clear scope boundaries (e.g., "Fix only the login form validation in src/auth.js. Touch nothing else.").
• Define stack constraints (e.g., "React 18, TypeScript strict, no external libraries, Tailwind only").
• Provide explicit file paths for IDE-integrated AIs (e.g., "Update handleLogin() in src/pages/Login.tsx only").
• Use the correct template for your tool (e.g., use File-Scope Template (Template G) for Cursor instead of GPT-style prose).
• Avoid pasting entire codebases; scope the context to only the relevant functions and files.

Use the Few-Shot template when the desired output format is easier to demonstrate through examples than to describe with text. This is highly effective for pattern replication and consistent structured output.

[Task instruction]

Here are examples of the exact format needed:

<examples>
  <example>
    <input>[example input 1]</input>
    <output>[example output 1]</output>
  </example>
  <example>
    <input>[example input 2]</input>
    <output>[example output 2]</output>
  </example>
</examples>

Now apply this exact pattern to: [actual input]

Use the CRISPE template for creative work, brand voice writing, and tasks where personality and iteration are important.

Capacity: [What capability or expertise the AI should have]
Role: [Specific persona to adopt]
Insight: [Key background insight that shapes the response]
Statement: [The core task or question]
Personality: [Tone and style — witty / authoritative / casual / sharp]
Experiment: [Request variants or alternatives to explore]

Use the Reference Image Editing template when modifying an existing image.

Important: Always instruct the user to attach the reference image to the tool before sending the prompt.

Tool-specific instructions:

• Midjourney: Use --cref [image URL] for character reference or --sref for style reference.
• DALL-E 3: Use the Edit endpoint (requires ChatGPT with image editing enabled).
• Stable Diffusion: Use img2img mode with denoising strength between 0.3-0.6.

Reference image: [attached / URL]
What to keep exactly the same: [list everything that must not change]
What to change: [specific edit only — be precise]
How much to change: [subtle / moderate / significant]
Style consistency: maintain the exact style, lighting, and mood of the reference
Negative prompt: [what to avoid introducing]

Use Chain of Thought (CoT) for logic-heavy tasks, math, debugging, and multi-factor analysis.

Warning: Do NOT use CoT instructions with reasoning models like o1, o3, or Claude extended thinking, as they reason internally and CoT instructions will degrade their output. Use CoT for standard reasoning models like Claude, GPT-4o, or Gemini.

[Task statement]

Before answering, think through this carefully:
<thinking>
1. What is the actual problem being asked?
2. What constraints must the solution respect?
3. What are the possible approaches?
4. Which approach is best and why?
</thinking>

Give your final answer in <answer> tags only.

For any prompt targeting agentic tools (Claude Code, Devin, Cursor, Windsurf, Cline, Bolt, SWE-agent, Manus, etc.) that involves filesystem, terminal, dependency, or database operations, you MUST append this notice:

This prompt is for an agentic tool with real system access. Review the scope locks, forbidden actions, and stop conditions before pasting. Confirm file paths, directories, and permissions match the actual project.

Before delivering a generated prompt, perform a final verification using the following criteria to ensure high success rates:

1. Tool Alignment: Verify the target tool is correctly identified and the prompt uses its specific syntax.
2. Constraint Primacy: Ensure the most critical constraints are located within the first 30% of the prompt.
3. Signal Strength: Use strong signal words (e.g., MUST instead of should, NEVER instead of avoid).
4. Authenticity: Remove all fabricated techniques.
5. Token Efficiency: Audit for load-bearing sentences, explicit formats, and bounded scopes; remove vague adjectives.
6. First-Attempt Success: Confirm the prompt is designed to produce the correct output on the first attempt without re-prompting.

Success Metric: The prompt must work on the first try with zero re-prompts required.