Full body text = references/full.md (약 7,000줄 — Read in full ❌). This SKILL.md is a thin index that does not duplicate content (structural prevention of the regression where the 2026-03 split version diverged from the flat baseline file and rotted — preventing dual representation staleness). Content edits always go only in references/full.md.
How to load (/prompt template load gate)
grep -n "<앵커 문자열>" skills/prompt-engineering-guide/references/full.md
→ 해당 offset 부분 Read
Section map (grep anchors — line numbers are reference values as-of 2026-07-29, anchor strings are the standard)
| Purpose | grep anchor | Location (reference) |
|---|---|---|
| Operating principles·terminology | ## 단일 스킬 운영 원칙 / ## 용어 해설 |
L16·L38 |
| Context Engineering | ## Context Engineering 원칙 적용 / ## Context Engineering 심층 통합 |
L98·L1446 |
| Expert priming | ## Expert Domain Priming |
L124 |
| Per-model strategy (GPT-5.2/Codex/5.5·Claude 4.5·Gemini 3·Veo) | ## 모델별 프롬프트 전략 |
L151~ |
| Purpose-specific blocks·level of detail | ## 목적별 추가 블록 / ## 상세도별 출력 지침 |
L1352·L1406 |
| Improvement workflow·quality | ## 프롬프트 개선 워크플로우 / ## 품질 체크리스트 |
L1509·L1656 |
| Agentic patterns | ## 에이전틱 워크플로우 패턴 |
L1711 |
| Intermediate structuring (storyboard·outline) | ## 중간 구조화 워크플로우 |
L1813 |
| Factcheck base template | LoopFactChecker / QuickFactCheck |
L5262·L5414 |
| Research base template | StructuredResearch_v1.0 |
L5445 |
| Unified appendix (merged from previously separate skills) | ## 통합 부록 |
L1947~ |
Related skills
- Image = ../image-prompt-guide/ (Gongnyang kit v4 original)
- Per-model details = ../claude-fable-5-prompt-strategies.md · ../gpt-5.6-prompt-enhancement/ · ../gemini-3.1-prompt-strategies/
- Expert DB = ../expert-domain-priming/ · Research = ../research-prompt-guide/