prompt-engineering-anthropic

@kevinjdurant

Use when the user asks to improve, rewrite, debug, audit, or optimize a prompt for Anthropic Claude models, especially Claude Opus 4.7 and Claude 4.6-family prompts. Also use for Claude prompt design involving response length, effort and adaptive thinking, tool use, progress updates, literal instruction scope, tone, subagents, interactive coding products, code review harnesses, computer use, clear instructions, context, examples, XML tags, roles, long-context prompting, model identity, output formatting, agentic systems, research, coding-agent guardrails, and vision tasks.