prompt-engineering

A reasoning mode where Claude dynamically decides whether and how deeply to think based on request complexity, replacing the older extended thinking mode that required manually setting a fixed token budget.

Prompting the model to write out intermediate reasoning steps before giving a final answer, rather than jumping straight to the answer, to improve accuracy on complex reasoning tasks.

Instead of describing in words what output you want, giving a few concrete input-output example pairs directly, letting Claude infer the style and format you're after from the examples themselves — often conveying detail more precisely than a written description can.

Breaking a complex task into a sequence of smaller prompts where each step's output feeds the next, trading a single mega-prompt for checkpoints you can inspect and correct along the way.

Setting Claude's role in a single system-prompt sentence to focus its tone and behavior — officially positioned as a short behavioral anchor, not an elaborate role-play persona.

An instruction layer set before the conversation begins, separate from user input, that defines how the AI behaves and what rules it follows throughout — and carries more weight than instructions a user gives mid-conversation.

A parameter controlling how much randomness Claude injects when choosing its next word, ranging 0.0–1.0 — lower values produce more consistent, predictable output, higher values produce more varied output, though even 0.0 doesn't guarantee fully identical results every time.