Prompt Engineering Overview

# 005. Prompt Engineering Overview (Claude Docs)

Anthropic, official docs, living page. EVIDENCE-BASED.
URL: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/overview (retrieved 2026-07-07)

## What it says

- Orders techniques by leverage, paraphrased: be clear and direct, use examples (multishot), let Claude think (chain of thought), use XML tags to structure, assign a role via system prompt, prefill responses, chain complex prompts.
- Treats prompt engineering as cheaper and faster than fine-tuning for most behavior shaping.
- Companion pages add long-context tips and Claude 4/5 era guidance (explicit instruction following, parallel tool use).

## Why it is canon here

It is the official baseline every prompting claim in this brain is measured against; [[Prompting Fable 5 One-Pager]] and [[Prompt Engineering Workflow for Fable 5]] distill it for the Fable 5 era.

## How this brain applies it

- Notes tag prompting claims EVIDENCE-BASED only when this page or its siblings state them.
- Adaptive-thinking guidance for Fable 5 (effort parameter, no raw chain of thought) is layered on top in [[Extended Thinking Budgets]].