Request: Prompt Engineering skill for Claude models

## Summary

There is no official skill for prompt engineering with Claude models (Opus 4.6, Sonnet 4.6, Haiku 4.5). This would be extremely valuable as a maintained, installable skill.

## The Gap

Anthropic has excellent prompt engineering documentation at [platform.claude.com](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices) and an [interactive tutorial](https://github.com/anthropics/prompt-eng-interactive-tutorial), but neither is packaged as an agent skill.

Community alternatives exist (prompt-master, prompt-architect, etc.) but they either target all LLMs generically or go stale when new models are released. None offer deep, Claude-specific guidance that stays current.

## What this skill would cover

- Model-specific prompting patterns for Opus, Sonnet, and Haiku
- When to use XML tags, chain of thought, extended thinking
- System vs user prompt best practices
- Token optimization per model tier
- Tool use and agentic prompting patterns
- Updated guidance when new models are released

## Why it matters

An official skill would stay current with model releases, be authoritative, and save users from relying on community skills that quickly become outdated. It would also complement the existing `claude-api` skill, which covers SDK usage but not prompt design.