Prompt Engineering

Prompt Engineering Skill

A Claude Code skill for **understanding, creating, improving, and debugging prompts**. It helps Claude recognize user intent, handle simple and complex requests appropriately, manage ambiguity, and verify results before considering a task complete.

✨ What It Does

This skill helps Claude:

• Understand the user's actual goal instead of only following literal wording
• Recognize simple, moderate, and complex tasks
• Extract requirements, constraints, context, and expected outcomes
• Create structured prompts for coding, debugging, research, design, and planning
• Improve vague or poorly structured prompts
• Detect contradictions and unsupported assumptions
• Avoid unnecessary questions and unnecessary work
• Research deeply when evidence is required
• Choose the smallest reliable solution
• Verify important work before reporting completion
• Recover when an approach fails or the task starts going in the wrong direction

🧠 Core Workflow

The skill follows an adaptive workflow:

Understand
    ↓
Classify
    ↓
Inspect
    ↓
Plan when necessary
    ↓
Execute
    ↓
Verify
    ↓
Communicate

Simple requests should remain simple. Complex requests receive deeper planning, context inspection, and verification.

📁 Structure

prompt-engineering/
├── SKILL.md
├── README.md
├── LICENSE
└── references/
    ├── prompt-patterns.md
    ├── prompt-evaluation.md
    └── prompt-debugging.md

Files

**`SKILL.md`** The main instructions that define how Claude should understand and handle prompts.

**`README.md`** Human-facing documentation explaining the purpose and structure of the skill.

**`prompt-patterns.md`** Reusable patterns for coding, debugging, research, design, architecture, release tasks, and general prompt construction.

**`prompt-evaluation.md`** Methods for evaluating prompt quality, checking requirements, detecting ambiguity, and improving reliability.

**`prompt-debugging.md`** Guidance for di