prompt-engineeringlisted

<!-- Version: 2026-05 | Claude 4.x (Opus 4.8, Sonnet 4.6, Haiku 4.5) Source: https://platform.claude.com/docs/build-with-claude/prompt-engineering/claude-4-best-practices --> # Prompt Engineering for Claude 4 Claude 4 models follow instructions precisely. Vague prompts get literal interpretations. Apply these patterns. ## Core Principles 1. **Be explicit** - Claude does exactly what you ask, not what you might mean 2. **Add context** - Explain WHY, not just WHAT (Claude generalizes from reasoning) 3. **Match examples to intent** - Claude treats examples as specifications 4. **Use XML tags** - Structure complex instructions with semantic tags 5. **Avoid "think"** - Use "consider", "evaluate", "analyze" instead (when thinking is disabled) --- ## Decision Tree 1: Writing Instructions ``` START: Writing an instruction │ ├─► Is the desired behavior obvious? │ NO → Add explicit detail │ Example: "Create a dashboard" → "Create a dashboard with charts, │ filters, and export functionality. Go beyond basics." │ ├─► Could Claude interpret this literally in an unhelpful way? │ YES → Add context explaining WHY │ Example: "NEVER use ellipses" → "Never use ellipses because │ text-to-speech can't pronounce them." │ └─► Do I want Claude to be proactive or conservative? PROACTIVE → "Implement changes rather than suggesting them" CONSERVATIVE → "Default to rec