skills.rest
apsulli
apsulliCOMMUNITY
Explore all skills by this author →
GitHub

prompt-master

Generates tool-specific prompts for LLMs, IDEs, image, video, and workflow AI systems.

Updated Mar 11, 2026
Search Tags:#prompt engineering#prompt optimization#llm#claude#midjourney#prompt generator#cursor
One-click install
npx skills add https://github.com/apsulli/gsd-extended --skill prompt-master-apsulli
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: prompt-master
Source: https://github.com/apsulli/gsd-extended/tree/main/.agent/skills/prompt-master
Command: npx skills add https://github.com/apsulli/gsd-extended --skill prompt-master-apsulli

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Writing prompts that work on the first attempt is hard: vague tasks, missing constraints, wrong syntax for the target tool, and fabricated reasoning techniques waste tokens and force repeated re-prompting. This Skill converts a rough idea into a single production-ready prompt optimized for the specific AI tool being used. ## Core Features & Use Cases - Tool-Specific Routing: Detects the target AI system (Claude, GPT-4o, o1/o3, Cursor, Copilot, Midjourney, DALL-E, Stable Diffusion, ComfyUI, Sora, ElevenLabs, Zapier, and more) and applies the correct prompt structure and syntax for that category. - Diagnostic Repair: Scans user-provided prompts for task, context, format, scope, reasoning, and agentic failure patterns, then fixes them silently. - Prompt Decompiler Mode: Breaks down, adapts, simplifies, or splits an existing prompt into a chain when the user pastes one for analysis. - Use Case: A user says "write me a Midjourney prompt for a product photo" — the Skill extracts intent, asks up to 3 clarifying questions if needed, and outputs a copyable prompt with comma-separated descriptors and parameters like --ar and --v 6, plus a token estimate and strategy note. ## Quick Start Ask the AI to write a prompt for your target tool, for example: "Help me write a prompt for Claude Code that refactors my authentication module."

Frequently Asked Questions about prompt-master

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a better prompt for ChatGPT or Claude?▼

State your rough idea and target tool, and the Skill extracts task, output format, and constraints before generating a single copyable prompt. It uses XML tags for Claude, explicit format locks, and numeric constraints instead of vague adjectives.

How to fix a prompt that keeps giving bad results?▼

Paste the failing prompt and it is scanned against diagnostic checklists covering vague task verbs, missing success criteria, absent format locks, and scope problems. Fixes are applied silently unless they change your original intent.

Should I add chain-of-thought instructions to o1 or DeepSeek-R1 prompts?▼

No. Reasoning-native models like o1, o3, and DeepSeek-R1 think internally, and explicit chain-of-thought scaffolding degrades their output. Give them short, clean instructions stating what you want, not how to reason.

Does this work with image generators like Midjourney and Stable Diffusion?▼

Yes. It detects generation versus reference-editing tasks and applies tool-specific syntax: comma-separated descriptors with parameters for Midjourney, weighted token syntax with mandatory negative prompts for Stable Diffusion, and prose for DALL-E 3.

What prompt techniques should be avoided in single-prompt execution?▼

Avoid Mixture of Experts, Tree of Thought, Graph of Thought, Universal Self-Consistency, and layered prompt chaining, since a single forward pass cannot actually perform them and the model fabricates the behavior. The Skill replaces these with natively reliable alternatives.

When should prompts for agentic AI tools include stop conditions?▼

Always. Agentic tools like Claude Code, Devin, and SWE-agent require starting state, target state, allowed and forbidden actions, and explicit stop conditions, because runaway loops are the biggest credit killer in agentic workflows.

Related Skills

Explore More Content & Communication Skills→
mr-daedaliummr-daedalium
107

office-hours

Guide structured design sessions to produce action-oriented design documents.

Community
Advanced
zachmayerzachmayer
2

prompt-evolution

Evolve LLM prompts through genetic algorithms and multi-model evaluation.

Community
Advanced
nidhinjsnidhinjs
12.2k

prompt-master

Generates optimized prompts for AI tools including LLMs, coding agents, and image generators.

Community
Advanced
RJB07RJB07

gemini

Answer user questions and generate summaries via Gemini CLI.

Community
Basic