# AI Usage Mastery

> Expert guidance and structured strategies for using SOTA AI models (Grok, Claude, GPT, Gemini, etc.) effectively. Triggers when users ask about prompt engineering, best practices for LLMs/SOTA models, how to use Grok or AI well, improving prompt quality, iterative workflows, advanced prompting techniques, or leveling up AI collaboration skills.

- Skill: `florencevision/ai-usage-mastery` (Agent Skill)
- Install (CLI): `npx skillmds@latest add florencevision/ai-usage-mastery`
- Raw SKILL.md: https://api.skillmd.com/api/skills/florencevision/ai-usage-mastery/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: florencevision (https://skillmd.com/u/florencevision)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/florencevision/ai-usage-mastery

---


# AI Usage Mastery

When this skill is active, follow these principles to deliver consistently excellent advice on mastering SOTA AI models. Use imperative language, concrete examples, and focus on actionable strategies.

## Core Principles to Teach
- Treat AI as a high-powered collaborator. Results depend on clear instructions, rich context, and deliberate iteration.
- Vague prompts produce vague outputs. Specificity, constraints, and examples are the highest-leverage improvements.
- Always iterate. The first response is a starting point.
- Combine techniques: role + CoT + few-shot + strict format.

## Recommended Prompt Anatomy
Structure every strong prompt with these layers:
1. **Role/Persona**: "You are a world-class [expert role]..."
2. **Context**: Background, audience, constraints, references.
3. **Task/Goal**: Precise outcome + success criteria.
4. **Reasoning Instructions**: "Think step by step" or "Explain your reasoning."
5. **Examples (Few-shot)**: 1-3 demonstrations when format or style matters.
6. **Output Format & Constraints**: Exact structure, length, JSON/markdown, etc.
7. **Quality Anchors**: "Be maximally truthful. Consider counterarguments."

## High-Impact Techniques
- **Chain-of-Thought (CoT)**: Show intermediate reasoning. Zero-shot or few-shot versions.
- **Few-Shot Prompting**: Provide input→output examples for consistency.
- **Role Assignment**: Instantly shifts expertise and depth.
- **Self-Critique**: Ask model to review and revise its answer.
- **Decomposition & Chaining**: Break tasks into sequential prompts.
- **Structured Output**: Always specify format (markdown, JSON, XML tags).
- **Tool Triggering (Grok)**: Explicitly say "Search X/web for latest..." or "Use DeepSearch..."

## Grok-Specific Mastery
- **Real-time strength**: Explicitly trigger X search, web search, DeepSearch for current events, sentiment, trends.
- **Image generation/editing**: Demand highly detailed prompts (subject, style, lighting, composition, references).
- **Document creation**: Direct creation of polished PDF, Word, PPTX with structure.
- **Tone control**: Request "unhedged honest take" or specific style.
- **Workflows**: Grok for real-time research/sentiment → other models for deep analysis → back to Grok for validation.

## Response Structure When Teaching
1. Golden rules + prompt template.
2. Core techniques with why + examples.
3. Grok-specific tactics + trigger phrases.
4. Practical workflow example.
5. Pitfalls + offer to build custom templates.

## Pitfalls to Highlight
- Overly long/ambiguous prompts.
- Assuming real-time knowledge without tool triggers.
- Skipping verification on facts.
- One-model-for-everything instead of strengths-based workflows.
- No iteration.

This skill ensures consistent, high-signal coaching on AI mastery.