Not in the top 1,000 by GitHub stars
treylom/prompt-engineering-skills
Comprehensive AI prompt engineering skills for Claude Code, ChatGPT (GPTs & Codex CLI plugin), and Gemini Gems. Model-specific patterns for Claude Opus 5.5 / Fable 5.1 (Opus 5 / Opus 4.8 legacy), GPT-6 Sol · Luna · Astra (GPT-5.6 Sol legacy), Gemini 3.1, Veo 3.1, Nano Banana 2, and gpt-image-2.
History
Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.
View weekly changes
instructions/Gems-Prompt-Generator/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/claude-4.7-prompt-strategies/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/expert-domain-priming/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/gemini-3.1-prompt-strategies/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/gpt-5.6-prompt-enhancement/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/image-prompt-guide/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Local scripts → After Local scripts
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Local scripts
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/prompt/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/prompt-engineering-guide/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/prompt-variation-guide/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
skills/research-prompt-guide/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
4dacb687a5a4 → 4dacb687a5a4
Diff text is unavailable under the recorded licenses. Line counts and the GitHub link remain available.
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 4dacb687a5a4
No earlier version was observed for comparison.
View on GitHubAdd to my skills
Verified commit 4dacb687a5a41c8e114e3de64e46814e54ed40bc
git clone --filter=blob:none --no-checkout https://github.com/treylom/prompt-engineering-skills.git .hotskills/treylom-prompt-engineering-skills && git -C .hotskills/treylom-prompt-engineering-skills checkout --detach 4dacb687a5a41c8e114e3de64e46814e54ed40bc
# Copy each selected skill directory from .hotskills/treylom-prompt-engineering-skills into .claude/skills/<skill-name>/193GitHub stars
0stars in 7 days
10skills in this repository
0commits in 30 days
Heat
0.3- Star growth, last 7 daysNot available
- Star growth, last 24 hoursNot available
- Rank movement, last 7 daysNot available
- Commits, last 30 daysNot available
- Accounts asking about it through MCP, last 7 daysNot available
- Hot votes this weekNot available
Heat never changes the order by stars, and sponsorship never touches it. How heat works
Stars over time
30-day change not yet available: not enough history.
Individual skills in this repository
- gems-prompt-generator
instructions/Gems-Prompt-Generator - claude-4-7-prompt-strategies
skills/claude-4.7-prompt-strategies - expert-domain-priming
skills/expert-domain-priming - gemini-3-1-prompt-strategies
skills/gemini-3.1-prompt-strategies - gpt-5-6-prompt-enhancement
skills/gpt-5.6-prompt-enhancement - image-prompt-guide
skills/image-prompt-guide - prompt
skills/prompt - prompt-engineering-guide
skills/prompt-engineering-guide - prompt-variation-guide
skills/prompt-variation-guide - research-prompt-guide
skills/research-prompt-guide
How it works
gems-prompt-generator
When your agent uses it
Use when needing Google Gems 프롬프트 생성기. Gems 페르소나용 시스템 프롬프트 자동 생성.
What it loads into context
Frontmatter
42 tokensAlways available to the agent
SKILL.md
759 tokensWhen the skill is activated
Reference documents
16,530 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Overview
- Sub-files
- Quick Navigation
Files
../../.claude-plugin/marketplace.jsonasset750 bytes../../.claude-plugin/plugin.jsonasset557 bytes01-mindset-constraints.mddoc5,004 bytes02-role-model-rankings.mddoc4,247 bytes03-element-expansion-expert-priming.mddoc3,334 bytes04-workflow-detection-structuring.mddoc7,228 bytes05-workflow-generation-output.mddoc10,746 bytes06-gemini-guide-json-structures.mddoc5,937 bytes07-references-final-reminder.mddoc3,524 bytes08-changelog.mddoc14,107 bytesSKILL.mdcore2,546 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubclaude-4-7-prompt-strategies
When your agent uses it
Use when the user explicitly targets Claude Opus 4.7 or earlier (구세대 first-class 지원) — legacy Claude 프롬프트 전략 레퍼런스의 라우팅 인덱스. 최신 모델(Opus 5·Fable 5·4.8·Sonnet 5)은 claude-fable-5-prompt-strategies.md가 기준 파일. 본문 = references/full.md.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
457 tokensWhen the skill is activated
Reference documents
10,754 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값)
Files
SKILL.mdcore1,269 bytesreferences/full.mddoc38,222 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubexpert-domain-priming
When your agent uses it
Use when a prompt needs an expert role assignment — 실존 전문가 지명과 전문 용어로 AI의 잠재 공간을 활성화하는 전문가 도메인 프라이밍 방법론 레퍼런스(/prompt 전문가 프롬프팅 대원칙의 근거 자료)의 라우팅 인덱스. 본문 = references/full.md.
What it loads into context
Frontmatter
126 tokensAlways available to the agent
SKILL.md
398 tokensWhen the skill is activated
Reference documents
9,247 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값)
Files
SKILL.mdcore1,282 bytesreferences/full.mddoc34,273 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubgemini-3-1-prompt-strategies
When your agent uses it
Use when composing or optimizing prompts targeting Google Gemini models (Gemini 3 Pro/Flash, 3.1) — Google 공식 문서 + 실전 팁 기반 Gemini 프롬프트 전략 레퍼런스의 라우팅 인덱스. 본문 = references/full.md.
What it loads into context
Frontmatter
103 tokensAlways available to the agent
SKILL.md
371 tokensWhen the skill is activated
Reference documents
9,182 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값)
Files
SKILL.mdcore1,101 bytesreferences/full.mddoc32,505 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubgpt-5-6-prompt-enhancement
When your agent uses it
Use when composing prompts targeting GPT-6 Sol (GPT 디폴트), GPT-6 Astra, GPT-6 Luna, GPT-5.6 Sol (직전 디폴트), legacy GPT-5.5/5.4/5.2 — lean outcome-first 7블록 규칙 + LEGACY 섹션(구 XML 스택)을 담은 GPT 프롬프트 전략 레퍼런스의 라우팅 인덱스. 본문 = references/full.md.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
662 tokensWhen the skill is activated
Reference documents
14,415 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-09-27 참고값)
Files
SKILL.mdcore1,768 bytesreferences/full.mddoc53,074 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubimage-prompt-guide
When your agent uses it
Use when generating AI image prompts — 원본 = 공냥 프롬프트 킷 v4(gpt-image-2 텍스트 프롬프트 계약 — 포맷 A/B·철칙·jsonl·검증기, skills/image-prompt-kit/ vendor 원문). 웹 UI(ChatGPT/Gemini) JSON 간이 경로는 legacy 부록. /prompt 커맨드가 이미지 타겟 감지 시 references/full.md 전체 로드.
What it loads into context
Frontmatter
157 tokensAlways available to the agent
SKILL.md
621 tokensWhen the skill is activated
Reference documents
15,042 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 구조 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값)
- 원문 vendor
Scripts executed separately
../image-prompt-kit/scripts/check_prompt.mjs
Files
SKILL.mdcore1,992 bytesreferences/full.mddoc48,729 bytes../image-prompt-kit/scripts/check_prompt.mjsruns21,641 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubprompt
When your agent uses it
Use when the user asks to create, refine, or optimize a prompt ('프롬프트 만들어/생성해/다듬어') — AI 프롬프트 생성기. 사용자의 아이디어를 모델별(Claude/GPT/Gemini/이미지) 최적화 프롬프트로 변환. Codex 포트 — Claude Code의 /prompt 명령과 동일 절차.
What it loads into context
Frontmatter
91 tokensAlways available to the agent
SKILL.md
1,271 tokensWhen the skill is activated
Reference documents
55,149 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Codex 이미지 목적 선행 게이트
- 🚨 산출 후 워크플로 — 이미지 경로 포함 전 목적 의무 (스킵 = 계약 위반)
Scripts executed separately
../../scripts/prompt-output-gate.mjs
Files
../../scripts/prompt-output-gate.mjsruns18,749 bytes../image-prompt-guide/references/full.mddoc48,729 bytesSKILL.mdcore4,484 bytesreferences/prompt-command.mddoc132,154 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubprompt-engineering-guide
When your agent uses it
Use when generating research/factcheck/image/video/slide prompts that need base templates (IFCN·StructuredResearch 등) — 단일 통합 AI 프롬프트 엔지니어링 레퍼런스의 라우팅 인덱스. 본문 전체는 references/full.md(대형 파일 — grep으로 위치 확인 후 부분 Read).
What it loads into context
Frontmatter
107 tokensAlways available to the agent
SKILL.md
841 tokensWhen the skill is activated
Reference documents
87,107 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 로드 방법 (/prompt 템플릿 로드 게이트)
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값, 앵커 문자열이 기준)
- 관련 스킬
Files
../claude-fable-5-prompt-strategies.mddoc44,469 bytes../expert-domain-priming/SKILL.mddoc1,282 bytes../gemini-3.1-prompt-strategies/SKILL.mddoc1,101 bytes../gpt-5.6-prompt-enhancement/SKILL.mddoc1,768 bytes../image-prompt-guide/SKILL.mddoc1,992 bytesSKILL.mdcore2,688 bytesreferences/full.mddoc260,481 bytes../research-prompt-guide/SKILL.mddoc1,410 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubprompt-variation-guide
When your agent uses it
Use when needing AI 프롬프트 변주 가이드. 다양한 프롬프팅 테크닉과 변형 전략 레퍼런스.
What it loads into context
Frontmatter
47 tokensAlways available to the agent
SKILL.md
1,503 tokensWhen the skill is activated
Reference documents
8,439 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Part 1: Universal Framework
- Part 3.3: Model Personality Mapping (핵심 선택 기준)
- Part 4: Quality Validation
- 요약
Files
SKILL.mdcore4,963 bytesreferences/examples.mddoc20,189 bytesreferences/techniques.mddoc8,404 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubresearch-prompt-guide
When your agent uses it
Use when composing factcheck or research prompts — IFCN 원칙 기반 검증 워크플로우와 일반 리서치 템플릿을 담은 레퍼런스의 라우팅 인덱스. 본문 = references/full.md.
What it loads into context
Frontmatter
118 tokensAlways available to the agent
SKILL.md
431 tokensWhen the skill is activated
Reference documents
77,931 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 섹션 맵 (grep 앵커 — 줄번호는 as-of 2026-07-29 참고값)
Files
../prompt-engineering-guide/references/full.mddoc260,481 bytesSKILL.mdcore1,410 bytesreferences/full.mddoc21,598 bytes
No excerpt is shown for this license. Read the source on GitHub.
View the complete source on GitHubSkills included
- gems-prompt-generator
- claude-4-7-prompt-strategies
- expert-domain-priming
- gemini-3-1-prompt-strategies
- gpt-5-6-prompt-enhancement
- image-prompt-guide
- prompt
- prompt-engineering-guide
- prompt-variation-guide
- research-prompt-guide
Repository facts
- License
- No license detected
- Created
- Dec 28, 2025
- Last pushed
- Sep 26, 2026
- Forks
- 33
- Open issues
- 0
- Project activity
- Active
Before you install
Network or broad shellAutomated signals, not a guarantee
Checked Nov 7, 2026
- No broad tool permissions
- No dangerous shell commands
- 178 network referencesin 13 files: prompt-engineering-guide/references/full.md:465 prompt/references/prompt-command.md:3 image-prompt-guide/references/full.md:12 and 10 more
- 1 obfuscation signalin 1 file: skills/image-prompt-kit/scripts/check_prompt.mjs:94
- No instruction overrides
- 2 executable scriptsin 2 files: scripts/prompt-output-gate.mjs:1 skills/image-prompt-kit/scripts/check_prompt.mjs:1
- No access to secrets
- No requests to vote in Hot
Show all 181 signals by file
.claude-plugin/marketplace.json
.claude-plugin/plugin.json
- Network accessline 5
"author": { "name": "tofukyung", "url": "https://github.com/treylom" - Network accessline 6
"name": "tofukyung", "url": "https://github.com/treylom" }, - Network accessline 7
"url": "https://github.com/treylom" - Network accessline 8
}, "homepage": "https://github.com/treylom/prompt-engineering-skills", "repository": "https://github.com/treylom/prompt-engineering-skills", - Network accessline 9
"homepage": "https://github.com/treylom/prompt-engineering-skills", - Network accessline 10
"repository": "https://github.com/treylom/prompt-engineering-skills",
instructions/Gems-Prompt-Generator/02-role-model-rankings.md
- Network accessline 22
> 출처: [LMArena Leaderboard](https://lmarena.ai) 기준 사용자 투표 순위 + 2026-09-27 모델 라인업 반영 (Claude 디폴트 = **Opus 5.5** · 최고난도 추론·장기 에이전트 = **Fable 5.1** / GPT 디폴트 = **GPT-6 Sol** · 최고 성능 = GPT-6 Astra · 대량·반복 = GPT-6 Luna). 신모델(2026-09-22 출시) 자리는 공
instructions/Gems-Prompt-Generator/05-workflow-generation-output.md
- Network accessline 100
> **다중 이미지 생성 시 추가 안내**: gemini에서 여러 장의 이미지를 생성할 경우, **'한 장씩 순차적으로 생성, 반드시 끝까지 다 생성해주세요'**도 함께 입력해주세요 > > **동영상 생성**: [Sora 2](https://sora.com) | [Veo 3.1 Flow](https://labs.google/fx/tools/flow) - Network accessline 102
> **동영상 생성**: [Sora 2](https://sora.com) | [Veo 3.1 Flow](https://labs.google/fx/tools/flow)
instructions/Gems-Prompt-Generator/07-references-final-reminder.md
- Network accessline 43
| 플랫폼 | 링크 | |--------|------| | **Sora 2** | https://sora.com | - Network accessline 44
|--------|------| | **Sora 2** | https://sora.com | | **Veo 3.1 (Flow)** | https://labs.google/fx/tools/flow | - Network accessline 45
| **Sora 2** | https://sora.com | - Network accessline 46
| **Veo 3.1 (Flow)** | https://labs.google/fx/tools/flow |
instructions/Gems-Prompt-Generator/SKILL.md
- Network accessline 9
> **Version**: 3.4.0 | **Updated**: 2026-09-27 > **Credits**: 이미지 프롬프트 가이드 - 공냥이(@specal1849) > **Model Rankings**: [LMArena Leaderboard](https://lmarena.ai) + 2026-09-27 모델 라인업 반영 (Claude 디폴트 = Opus 5.5 · 최고난도 = Fable 5.1 / GPT 디폴트 = GPT-6 - Network accessline 10
> **Credits**: 이미지 프롬프트 가이드 - 공냥이(@specal1849) > **Model Rankings**: [LMArena Leaderboard](https://lmarena.ai) + 2026-09-27 모델 라인업 반영 (Claude 디폴트 = Opus 5.5 · 최고난도 = Fable 5.1 / GPT 디폴트 = GPT-6 Sol — 상세 = 02) > **Optimized for**: Gemini 3, - Network accessline 11
> **Model Rankings**: [LMArena Leaderboard](https://lmarena.ai) + 2026-09-27 모델 라인업 반영 (Claude 디폴트 = Opus 5.5 · 최고난도 = Fable 5.1 / GPT 디폴트 = GPT-6 Sol — 상세 = 02)
scripts/prompt-output-gate.mjs
- Executable scriptline 1
#!/usr/bin/env node
skills/claude-4.7-prompt-strategies/references/full.md
- Network accessline 9
> **⚠<U+FE0F> 최신 모델 (Opus 4.8 · Fable 5) = `claude-fable-5-prompt-strategies.md` 참조** — 2026-06-10부터 Claude 디폴트는 Opus 4.8. 본 파일은 4.7 이하 구세대 first-class 지원용으로 유지. > **Version**: 3.2.0 | **Updated**: 2026-04-30 > **Source**: Anthropic 공식 문서 ([platfo - Network accessline 10
> **Version**: 3.2.0 | **Updated**: 2026-04-30 > **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Migratio - Network accessline 11
> **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Migration](https://platform.claude.com/docs/en/docs/abo - Network accessline 127
| 갑자기 emoji/축하어 사라짐 | 4.7 직접적 어조 | 정상 — 필요 시 페르소나 블록에서 명시적으로 톤 지정 | | LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 - Network accessline 128
| LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 | | 단순 요청에도 thinking 폭주 | adaptive 자율 판단 + effort 과다 | "Extended th - Network accessline 129
| 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 |
skills/claude-fable-5-prompt-strategies.md
- Network accessline 20
|------|------| | **미지정·일반 작업 (디폴트)** | **Opus 5.5** (`claude-opus-5-5`) — effort 기본 `medium`(명시 권장) · thinking 항상 켜짐 · 1M context (Part 2.7) | | 어려운 추론·장기 에이전트 작업 | **Fable 5.1** — 공식 모델 개요는 Opus 5.5 를 출발점으로 둔다: *"If you're unsure which mo - Network accessline 21
| **미지정·일반 작업 (디폴트)** | **Opus 5.5** (`claude-opus-5-5`) — effort 기본 `medium`(명시 권장) · thinking 항상 켜짐 · 1M context (Part 2.7) | | 어려운 추론·장기 에이전트 작업 | **Fable 5.1** — 공식 모델 개요는 Opus 5.5 를 출발점으로 둔다: *"If you're unsure which model to use, star - Network accessline 22
| 어려운 추론·장기 에이전트 작업 | **Fable 5.1** — 공식 모델 개요는 Opus 5.5 를 출발점으로 둔다: *"If you're unsure which model to use, start with Claude Opus 5.5 for most workloads."* · Fable 5.1 은 Opus 5.5 모델 개요 비교표에서 Latency «Slower»(Opus 5.5 = «Moderate») · 가격 입력/ - Network accessline 52
> **공식 출처**: [Prompting Claude Opus 5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5) · [What's new in Opus 5](https://platform.claude.com/docs/en/about-claude/models/whats-new-opus-5) · [M - Network accessline 174
> **공식 출처**: [Prompting Claude Fable 5.1](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-1) · [What's new in Claude Fable 5.1](https://platform.claude.com/docs/en/models/fable-5-1/whats-new - Network accessline 232
> **공식 출처**: [Prompting Claude Opus 5.5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5-5) · [What's new in Claude Opus 5.5](https://platform.claude.com/docs/en/models/opus-5-5/whats-new-opu - Network accessline 233
> 공식 모델 선택 문장: *"If you're unsure which model to use, start with Claude Opus 5.5 for most workloads."* · Fable 5.1 은 Opus 5.5 모델 개요 비교표에서 Latency «Slower»(Opus 5.5 = «Moderate») · 가격 입력/출력 100만 토큰당 $10 / $50(Opus 5.5 = $4 / $20) · 기본 effort
skills/expert-domain-priming/references/full.md
- Network accessline 11
source_credits: - name: "최승준 소장님, 노정석" context: "'Master Class - 원리를 생각하는 프롬프팅' — https://erucipe.notion.site/2e3d5c9e7e598053ae93e9ff5951dcaa , https://www.youtube.com/watch?v=Q43tbLNx21A&t=1107s" - Network accessline 12
- name: "최승준 소장님, 노정석" context: "'Master Class - 원리를 생각하는 프롬프팅' — https://erucipe.notion.site/2e3d5c9e7e598053ae93e9ff5951dcaa , https://www.youtube.com/watch?v=Q43tbLNx21A&t=1107s" - name: 공냥이(@specal1849) - Network accessline 13
context: "'Master Class - 원리를 생각하는 프롬프팅' — https://erucipe.notion.site/2e3d5c9e7e598053ae93e9ff5951dcaa , https://www.youtube.com/watch?v=Q43tbLNx21A&t=1107s" - Network accessline 15
context: "프롬프트 쿠튀르: 전문가의 AI 활용법" - name: erucipe (Notion A-Z) url: https://erucipe.notion.site/A-Z-2e3d5c9e7e5980c4a593f1994582f1ab - Network accessline 16
- name: erucipe (Notion A-Z) url: https://erucipe.notion.site/A-Z-2e3d5c9e7e5980c4a593f1994582f1ab --- - Network accessline 17
url: https://erucipe.notion.site/A-Z-2e3d5c9e7e5980c4a593f1994582f1ab
skills/gpt-5.6-prompt-enhancement/references/full.md
- Network accessline 9
> **Version**: 1.4.0 | **Updated**: 2026-09-27 (GPT-6 Sol·Luna 절 신설 — GPT 디폴트를 GPT-6 Sol 로 전환, `none` 지원·Chat Completions 함수 호출 조건·샘플링 매개변수 제거·출시 공지 기준 문체·GPT-5.6 Sol → GPT-6 Sol/Luna 마이그레이션 체크. 공식 문서 대조.) | 이전: 1.3.0 · 2026-09-06 > **Sourc - Network accessline 10
> **Source**: [Prompting guidance for GPT-5.6 Sol](https://developers.openai.com/api/docs/guides/prompt-guidance-gpt-5p6) · [Using GPT-6](https://developers.openai.com/api/docs/guides/latest-model) (GPT-6 패밀리 가이드 — 구 표기 «Using GPT-6 Astra») - Network accessline 279
> **공식 출처**: [Using GPT-6 Astra](https://developers.openai.com/api/docs/guides/latest-model) — 접근일 2026-09-05 - Network accessline 333
> **공식 출처** (접근일 2026-09-27): [Using GPT-6](https://developers.openai.com/api/docs/guides/latest-model) (GPT-6 패밀리 가이드) · 모델 카드 [gpt-6-sol](https://developers.openai.com/api/docs/models/gpt-6-sol.md) · [gpt-6-luna](https://developers.openai - Network accessline 455
- GPT-5.6 Sol 용으로 쓰던 문체·길이 지시("간결하게" 등)는 GPT-6 Sol·Luna 에서 그대로 필요한지 다시 잰다 — 배포 체크리스트의 "check whether broad instructions like \"Be concise\" still help" 와 같은 방향. 기본 상세도는 `text.verbosity`, 프롬프트는 필수 내용·구조·길이만. - 공지의 "verbosity sweeps" 는 사실성 평가 - Network accessline 457
- **여러 모델이 함께 쓰는 지시문 주의.** 공식 Astra 블로그 글(https://developers.openai.com/blog/rethinking-skills-and-prompts-for-gpt-6-astra.md) 원문: "Guidance that helps Sol or Luna may overconstrain GPT-6 Astra, so consider which models will use the instruc - Network accessline 925
- [Using GPT-6 (공식, GPT-6 패밀리 가이드)](https://developers.openai.com/api/docs/guides/latest-model) · 모델 카드 [gpt-6-sol](https://developers.openai.com/api/docs/models/gpt-6-sol.md) · [gpt-6-luna](https://developers.openai.com/api/docs/models/gpt - Network accessline 926
- [GPT-5.6 Sol Prompt Guidance (공식)](https://developers.openai.com/api/docs/guides/prompt-guidance-gpt-5p6) - Network accessline 927
- [GPT-5.5 Prompt Guidance (공식)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5)
skills/image-prompt-guide/references/full.md
- Network accessline 12
source_credits: - name: 공냥이(@specal1849) url_kit: https://github.com/gongnyang/gongnyang-prompt-kit - Network accessline 12
source_credits: - name: 공냥이(@specal1849) url_kit: https://github.com/gongnyang/gongnyang-prompt-kit - Network accessline 13
- name: 공냥이(@specal1849) url_kit: https://github.com/gongnyang/gongnyang-prompt-kit url_slides: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 13
- name: 공냥이(@specal1849) url_kit: https://github.com/gongnyang/gongnyang-prompt-kit url_slides: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 14
url_kit: https://github.com/gongnyang/gongnyang-prompt-kit - Network accessline 14
url_kit: https://github.com/gongnyang/gongnyang-prompt-kit - Network accessline 15
url_slides: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 15
url_slides: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 16
url_notion: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 16
url_notion: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 21
> **자료 출처**: 원본 계약은 공냥이(@specal1849)님의 [공냥 프롬프트 킷 v4](https://github.com/gongnyang/gongnyang-prompt-kit)(MIT — 원문 사본 `skills/image-prompt-kit/`)를 따릅니다. 후반부 일반 이론·웹 UI 경로는 같은 저자의 "이미지 프롬프트 101" 슬라이드와 "PRO 이미지 프롬프트" Notion 문서 기반입니다. - Network accessline 21
> **자료 출처**: 원본 계약은 공냥이(@specal1849)님의 [공냥 프롬프트 킷 v4](https://github.com/gongnyang/gongnyang-prompt-kit)(MIT — 원문 사본 `skills/image-prompt-kit/`)를 따릅니다. 후반부 일반 이론·웹 UI 경로는 같은 저자의 "이미지 프롬프트 101" 슬라이드와 "PRO 이미지 프롬프트" Notion 문서 기반입니다. - Network accessline 882
> **핵심 원칙**: 비영어 대사를 그대로만 넣으면 표정·억양이 어색해진다. 괄호 안에 **언어 표시 + 영어로 작성한 감정/말투 묘사**를 함께 적으면 자연스러워진다. > > 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모 - Network accessline 882
> **핵심 원칙**: 비영어 대사를 그대로만 넣으면 표정·억양이 어색해진다. 괄호 안에 **언어 표시 + 영어로 작성한 감정/말투 묘사**를 함께 적으면 자연스러워진다. > > 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모 - Network accessline 884
> 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모델 전반에 일반화 가능. - Network accessline 884
> 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모델 전반에 일반화 가능. - Network accessline 985
- **이미지 프롬프트 101 슬라이드**: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 985
- **이미지 프롬프트 101 슬라이드**: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 986
- **PRO 이미지 프롬프트 Notion**: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 986
- **PRO 이미지 프롬프트 Notion**: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 987
- **나노바나나pro 마스터하기**: https://www.threads.com/@specal1849/post/DRYW09dEtUi - Network accessline 987
- **나노바나나pro 마스터하기**: https://www.threads.com/@specal1849/post/DRYW09dEtUi - Network accessline 988
- **이미지 프롬프트 심화이론 - JSON vs YAML**: https://www.threads.com/@specal1849/post/DRp8XdFE2VH - Network accessline 988
- **이미지 프롬프트 심화이론 - JSON vs YAML**: https://www.threads.com/@specal1849/post/DRp8XdFE2VH - Network accessline 1083
- **안티패턴 4건 추가**: 한국어 메타·메타 분리·서술문 혼합·언어 누락 - **체크리스트 항목 추가**: "비영어 대사 시 영어 감정/말투 메타 지시 포함" - **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01) - Network accessline 1083
- **안티패턴 4건 추가**: 한국어 메타·메타 분리·서술문 혼합·언어 누락 - **체크리스트 항목 추가**: "비영어 대사 시 영어 감정/말투 메타 지시 포함" - **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01) - Network accessline 1084
- **체크리스트 항목 추가**: "비영어 대사 시 영어 감정/말투 메타 지시 포함" - **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01) **Changes v1.9.0**: - Network accessline 1084
- **체크리스트 항목 추가**: "비영어 대사 시 영어 감정/말투 메타 지시 포함" - **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01) **Changes v1.9.0**: - Network accessline 1085
- **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01) - Network accessline 1085
- **출처**: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) Threads 포스트 (Seedance 한국어 대사 검증, 2026-05-01)
skills/image-prompt-kit/scripts/check_prompt.mjs
skills/prompt-engineering-guide/references/full.md
- Network accessline 465
참조: [apply_patch 구현](https://github.com/openai/openai-cookbook/tree/main/examples/gpt-5/apply_patch.py) - Network accessline 465
참조: [apply_patch 구현](https://github.com/openai/openai-cookbook/tree/main/examples/gpt-5/apply_patch.py) - Network accessline 562
> **공식 출처**: [Prompt guidance for GPT-5.5](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) (2026-04) - Network accessline 562
> **공식 출처**: [Prompt guidance for GPT-5.5](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) (2026-04) - Network accessline 1780
| 문서 | URL | 주요 내용 | |------|-----|----------| | GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | - Network accessline 1780
| 문서 | URL | 주요 내용 | |------|-----|----------| | GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | - Network accessline 1781
|------|-----|----------| | GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | | GPT-5.4 Prompt Guidance | — | Output Contract, Agentic Patterns, Eagerness Control | - Network accessline 1781
|------|-----|----------| | GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | | GPT-5.4 Prompt Guidance | — | Output Contract, Agentic Patterns, Eagerness Control | - Network accessline 1782
| GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | - Network accessline 1782
| GPT-5.2 Prompting Guide | https://cookbook.openai.com/examples/gpt-5-2_prompting_guide | reasoning_effort, XML 구조, 장황함 제어 | - Network accessline 1783
| GPT-5.4 Prompt Guidance | — | Output Contract, Agentic Patterns, Eagerness Control | | GPT-5.2-Codex Prompting Guide | https://cookbook.openai.com/examples/gpt-5-codex_prompting_guide | 코딩 에이전트 전용 최적화, Anti-Prompting | | Introducing GPT-5 - Network accessline 1783
| GPT-5.4 Prompt Guidance | — | Output Contract, Agentic Patterns, Eagerness Control | | GPT-5.2-Codex Prompting Guide | https://cookbook.openai.com/examples/gpt-5-codex_prompting_guide | 코딩 에이전트 전용 최적화, Anti-Prompting | | Introducing GPT-5 - Network accessline 1784
| GPT-5.2-Codex Prompting Guide | https://cookbook.openai.com/examples/gpt-5-codex_prompting_guide | 코딩 에이전트 전용 최적화, Anti-Prompting | - Network accessline 1784
| GPT-5.2-Codex Prompting Guide | https://cookbook.openai.com/examples/gpt-5-codex_prompting_guide | 코딩 에이전트 전용 최적화, Anti-Prompting | - Network accessline 1785
| Introducing GPT-5.2-Codex | https://openai.com/index/introducing-gpt-5-2-codex/ | Codex 모델 소개 및 특성 | - Network accessline 1785
| Introducing GPT-5.2-Codex | https://openai.com/index/introducing-gpt-5-2-codex/ | Codex 모델 소개 및 특성 | - Network accessline 1789
| 문서 | URL | 주요 내용 | |------|-----|----------| | Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | - Network accessline 1789
| 문서 | URL | 주요 내용 | |------|-----|----------| | Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | - Network accessline 1790
|------|-----|----------| | Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | | Claude 4.6 Prompt Guide | — | Adaptive Thinking, Effort Parameter, - Network accessline 1790
|------|-----|----------| | Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | | Claude 4.6 Prompt Guide | — | Adaptive Thinking, Effort Parameter, - Network accessline 1791
| Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | - Network accessline 1791
| Claude 4.5 What's New | https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-5 | Claude 4.5 모델 변경사항, Extended Thinking | - Network accessline 1793
| Claude 4 Best Practices | https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-4-best-practices | 명시적 지시, 병렬 도구 호출, 프론트엔드 가이드라인 | - Network accessline 1793
| Claude 4 Best Practices | https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-4-best-practices | 명시적 지시, 병렬 도구 호출, 프론트엔드 가이드라인 | - Network accessline 1797
| 문서 | URL | 주요 내용 | |------|-----|----------| | Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | - Network accessline 1797
| 문서 | URL | 주요 내용 | |------|-----|----------| | Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | - Network accessline 1798
|------|-----|----------| | Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | | Image Generation (Nano Banana) | https://ai.google.dev/gemini-api/docs/image-generation?hl=ko | 이미지 생성 API, 프 - Network accessline 1798
|------|-----|----------| | Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | | Image Generation (Nano Banana) | https://ai.google.dev/gemini-api/docs/image-generation?hl=ko | 이미지 생성 API, 프 - Network accessline 1799
| Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | - Network accessline 1799
| Veo API Documentation | https://ai.google.dev/gemini-api/docs/video?hl=ko | 동영상 생성 API, 확장, 참조 이미지 | - Network accessline 1800
| Image Generation (Nano Banana) | https://ai.google.dev/gemini-api/docs/image-generation?hl=ko | 이미지 생성 API, 프롬프트 구조 | - Network accessline 1800
| Image Generation (Nano Banana) | https://ai.google.dev/gemini-api/docs/image-generation?hl=ko | 이미지 생성 API, 프롬프트 구조 | - Network accessline 1957
> 🆕 **GPT 기본 모델 = GPT-6 Sol** (2026-09-22~ · 직전 기본 = GPT-5.6 Sol): 최신 규칙은 `skills/gpt-5.6-prompt-enhancement/references/full.md`(GPT-6 절 · 5.6 본문) 참조. 아래 5.5 내용은 legacy 계보 보존. > **Version**: 1.1.0 | **Updated**: 2026-04-30 > **Source**: [Op - Network accessline 1957
> 🆕 **GPT 기본 모델 = GPT-6 Sol** (2026-09-22~ · 직전 기본 = GPT-5.6 Sol): 최신 규칙은 `skills/gpt-5.6-prompt-enhancement/references/full.md`(GPT-6 절 · 5.6 본문) 참조. 아래 5.5 내용은 legacy 계보 보존. > **Version**: 1.1.0 | **Updated**: 2026-04-30 > **Source**: [Op - Network accessline 1958
> **Version**: 1.1.0 | **Updated**: 2026-04-30 > **Source**: [OpenAI GPT-5.5 Prompt Guidance (2026-04)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) > **Scope**: GPT 5.x 구세대 통합 — outcome-first markdown(5.5) + - Network accessline 1958
> **Version**: 1.1.0 | **Updated**: 2026-04-30 > **Source**: [OpenAI GPT-5.5 Prompt Guidance (2026-04)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) > **Scope**: GPT 5.x 구세대 통합 — outcome-first markdown(5.5) + - Network accessline 1959
> **Source**: [OpenAI GPT-5.5 Prompt Guidance (2026-04)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) - Network accessline 1959
> **Source**: [OpenAI GPT-5.5 Prompt Guidance (2026-04)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) - Network accessline 2405
- [GPT-5.5 Prompt Guidance (공식)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) - Network accessline 2405
- [GPT-5.5 Prompt Guidance (공식)](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) - Network accessline 2414
> 🆕 **2026-09-27 갱신 — Claude 기본 모델 = Claude Opus 5.5** (`claude-opus-5-5`, 2026-09-22~). 공식 모델 안내: "If you're unsure which model to use, start with Claude Opus 5.5 for most workloads." · Fable 5.1 은 Opus 5.5 모델 개요 비교표에서 Latency «Slower»(Opu - Network accessline 2414
> 🆕 **2026-09-27 갱신 — Claude 기본 모델 = Claude Opus 5.5** (`claude-opus-5-5`, 2026-09-22~). 공식 모델 안내: "If you're unsure which model to use, start with Claude Opus 5.5 for most workloads." · Fable 5.1 은 Opus 5.5 모델 개요 비교표에서 Latency «Slower»(Opu - Network accessline 2416
> **Updated**: 2026-07-21 | **Source**: [Prompting Claude Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-4-8) + [Prompting Claude Fable 5](https://platform.claude.com/docs/en/build-w - Network accessline 2416
> **Updated**: 2026-07-21 | **Source**: [Prompting Claude Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-4-8) + [Prompting Claude Fable 5](https://platform.claude.com/docs/en/build-w - Network accessline 2443
> **Version**: 3.2.0 | **Updated**: 2026-04-30 | ⚠<U+FE0F> **구세대 참고 문서** — 현행(Opus 5.5 디폴트·Fable 5.1 · Opus 5·Opus 4.8·Fable 5·Sonnet 5)은 위 "Claude 현행 모델 전략" 섹션 참조 > **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](http - Network accessline 2443
> **Version**: 3.2.0 | **Updated**: 2026-04-30 | ⚠<U+FE0F> **구세대 참고 문서** — 현행(Opus 5.5 디폴트·Fable 5.1 · Opus 5·Opus 4.8·Fable 5·Sonnet 5)은 위 "Claude 현행 모델 전략" 섹션 참조 > **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](http - Network accessline 2444
> **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Migration](https://platform.claude.com/docs/en/docs/abo - Network accessline 2444
> **Source**: Anthropic 공식 문서 ([platform.claude.com — Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Migration](https://platform.claude.com/docs/en/docs/abo - Network accessline 2560
| 갑자기 emoji/축하어 사라짐 | 4.7 직접적 어조 | 정상 — 필요 시 페르소나 블록에서 명시적으로 톤 지정 | | LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 - Network accessline 2560
| 갑자기 emoji/축하어 사라짐 | 4.7 직접적 어조 | 정상 — 필요 시 페르소나 블록에서 명시적으로 톤 지정 | | LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 - Network accessline 2561
| LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 | | 단순 요청에도 thinking 폭주 | adaptive 자율 판단 + effort 과다 | "Extended th - Network accessline 2561
| LaTeX 안 나옴 | 4.7 기본 미사용 | `"Format math expressions in LaTeX"` 명시 | | 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 | | 단순 요청에도 thinking 폭주 | adaptive 자율 판단 + effort 과다 | "Extended th - Network accessline 2562
| 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 | - Network accessline 2562
| 사이버 보안 요청 거부 | 4.7 세이프가드 강화 | 정상 사용처라면 [Cyber Verification Program](https://platform.claude.com/) 신청 | - Network accessline 5063
> **핵심 원칙**: 비영어 대사를 그대로만 넣으면 표정·억양이 어색해진다. 괄호 안에 **언어 표시 + 영어로 작성한 감정/말투 묘사**를 함께 적으면 자연스러워진다. > > 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모 - Network accessline 5063
> **핵심 원칙**: 비영어 대사를 그대로만 넣으면 표정·억양이 어색해진다. 괄호 안에 **언어 표시 + 영어로 작성한 감정/말투 묘사**를 함께 적으면 자연스러워진다. > > 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모 - Network accessline 5065
> 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모델 전반에 일반화 가능. - Network accessline 5065
> 출처: [@itsyun_ai](https://www.threads.com/@itsyun_ai/post/DXyxUwXghcR) (Seedance 한국어 대사 검증, 2026-05-01). 시나리오 작법(stage direction) 컨벤션이라 동영상 모델 전반에 일반화 가능. - Network accessline 5166
- **이미지 프롬프트 101 슬라이드**: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 5166
- **이미지 프롬프트 101 슬라이드**: https://docs.google.com/presentation/d/1rPQVnbu1INJyUAqCvMA7dkO2WzJpD4Q9q_UXq9RH2GU/edit - Network accessline 5167
- **PRO 이미지 프롬프트 Notion**: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 5167
- **PRO 이미지 프롬프트 Notion**: https://fascinated-alley-b43.notion.site/PRO-2b1861d1faaf80b8bf7ef4093827f59b - Network accessline 5168
- **나노바나나pro 마스터하기**: https://www.threads.com/@specal1849/post/DRYW09dEtUi - Network accessline 5168
- **나노바나나pro 마스터하기**: https://www.threads.com/@specal1849/post/DRYW09dEtUi - Network accessline 5169
- **이미지 프롬프트 심화이론 - JSON vs YAML**: https://www.threads.com/@specal1849/post/DRp8XdFE2VH - Network accessline 5169
- **이미지 프롬프트 심화이론 - JSON vs YAML**: https://www.threads.com/@specal1849/post/DRp8XdFE2VH
skills/prompt/references/prompt-command.md
- Network accessline 3
> **Version**: 3.4.0 | **Updated**: 2026-09-27 — 🆕 **v3.4.0: Claude 디폴트 = Opus 5.5 · GPT 디폴트 = GPT-6 Sol**(2026-09-22 출시 · 공식 문서 대조, 접근 2026-09-27 — 머리 소스 줄·Claude/GPT 라우팅 표·batch 1번째 토큰·모델별 필수 블록·AskUserQuestion 모델 옵션 갱신). 이전: v3.3.2: 사용자 - Network accessline 4
> **Model Rankings**: [LMArena Leaderboard](https://lmarena.ai) (2026년 3월 기준) - Network accessline 5
> **이미지 프롬프트 소스 (원본 킷)**: **공냥 프롬프트 킷 v4** — [gongnyang/gongnyang-prompt-kit](https://github.com/gongnyang/gongnyang-prompt-kit) (MIT, 원문 vendor `skills/image-prompt-kit/` · 계약 전사 `skills/image-prompt-guide/references/full.md` §0·§K). 보조: [ - Network accessline 6
> **Opus 5.5 공식 소스 (현행 디폴트, 2026-09-22 출시 · `claude-opus-5-5`)**: [Prompting Claude Opus 5.5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5-5) + [What's new in Claude Opus 5.5](https://plat - Network accessline 7
> **Fable 5.1 공식 소스 (최고난도 추론·장기 에이전트)**: [Prompting Claude Fable 5.1](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-1) — 상세: `skills/claude-fable-5-prompt-strategies.md` Part 2.6 > **Opus - Network accessline 8
> **Opus 5 공식 소스 (직전 디폴트)**: [Prompting Claude Opus 5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5) + [What's new in Opus 5](https://platform.claude.com/docs/en/about-claude/models/whats- - Network accessline 9
> **Opus 4.8 / Fable 5 공식 소스 (구세대 first-class)**: [Prompting Claude Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-4-8) + [Prompting Claude Fable 5](https://platform.claude.com/docs/ - Network accessline 10
> **Opus 4.7 / 4.6 공식 소스 (구세대 first-class)**: [platform.claude.com/docs — Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Adaptive thinking](https://platform - Network accessline 11
> **GPT 공식 가이드 (디폴트 = GPT-6 Sol, 2026-09-22 출시 · `gpt-6-sol`)**: [Using GPT-6](https://developers.openai.com/api/docs/guides/latest-model) (접근 2026-09-27) — GPT-6 패밀리 가이드 1페이지(GPT-6 Sol·GPT-6 Luna 전용 프롬프트 가이드 = 확인 범위 내 미발견). 용도 원문: *"Use gp - Network accessline 12
> **GPT-5.6 공식 가이드 (직전 디폴트 = GPT-5.6 Sol, 2026-07)**: [Prompting guidance for GPT-5.6](https://developers.openai.com/api/docs/guides/prompt-guidance-gpt-5p6) — lean outcome-first. 상세: `skills/gpt-5.6-prompt-enhancement/references/full.md`. - Network accessline 253
> 출처: [LMArena Leaderboard](https://lmarena.ai) 기준 사용자 투표 순위 + 2026-09-27 모델 라인업 반영 (Claude 디폴트 = **Opus 5.5** · 최고난도 추론·장기 에이전트 = **Fable 5.1** / GPT 디폴트 = **GPT-6 Sol** · 최고 성능 = GPT-6 Astra · 대량·반복 = GPT-6 Luna). 신모델(2026-09-22 출시) 자리는 공 - Network accessline 360
> **출처 (신)**: [Prompting Claude Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-4-8) + [Prompting Claude Fable 5](https://platform.claude.com/docs/en/build-with-claude/prompt-engineer - Network accessline 361
> **출처 (구)**: [Claude 4 best practices](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Adaptive thinking](https://platform.claude.com/docs/en/docs/build-with-claude/adaptive-thinkin - Network accessline 477
| 구 prefill 용도 | 신 대안 | |----------------|--------| | 포맷 강제 (`{"start": `로 시작) | [Structured Outputs](https://platform.claude.com/docs/en/docs/build-with-claude/structured-outputs) 또는 `output_config.format` | - Network accessline 478
|----------------|--------| | 포맷 강제 (`{"start": `로 시작) | [Structured Outputs](https://platform.claude.com/docs/en/docs/build-with-claude/structured-outputs) 또는 `output_config.format` | | "Here is the summary:\n" preamble 제거 | system: `"Resp - Network accessline 479
| 포맷 강제 (`{"start": `로 시작) | [Structured Outputs](https://platform.claude.com/docs/en/docs/build-with-claude/structured-outputs) 또는 `output_config.format` | - Network accessline 512
> 🆕 **GPT 기본 모델 = GPT-6 Sol** (2026-09-22 출시): 최신 규칙은 `skills/gpt-5.6-prompt-enhancement/references/full.md` «GPT-6 Sol · Luna» 절 참조. 직전 디폴트 GPT-5.6 Sol(2026-07, lean outcome-first) = 같은 파일 5.6 본문. 아래 5.5 내용은 legacy 계보 보존. > **공식 가이드**: [GP - Network accessline 513
> **공식 가이드**: [GPT-5.5 Prompt Guidance](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) (2026-04 발표) - Network accessline 1580
> > 🎬 **동영상 생성**: 위 코드를 복사하여 아래 링크에서 사용하세요. > - **Veo 3.1 (Flow)**: https://labs.google/fx/tools/flow - Network accessline 1581
> 🎬 **동영상 생성**: 위 코드를 복사하여 아래 링크에서 사용하세요. > - **Veo 3.1 (Flow)**: https://labs.google/fx/tools/flow > - **Sora 2**: https://sora.com - Network accessline 1582
> - **Veo 3.1 (Flow)**: https://labs.google/fx/tools/flow - Network accessline 1583
> - **Sora 2**: https://sora.com - Network accessline 1831
| 플랫폼 | 링크 | |--------|------| | **Sora 2** | https://sora.com | - Network accessline 1832
|--------|------| | **Sora 2** | https://sora.com | | **Veo 3.1 (Flow)** | https://labs.google/fx/tools/flow | - Network accessline 1833
| **Sora 2** | https://sora.com | - Network accessline 1834
| **Veo 3.1 (Flow)** | https://labs.google/fx/tools/flow | - Network accessline 1914
- **Changes v2.9.0**: ①템플릿 로드 게이트 전역 승격 ②Opus 4.8·Fable 5 공식 문서 재fetch 전수 대조 - **Changes v2.8.0** (2026-06-10): - **[MAJOR] Claude 디폴트 = Opus 4.8 격상**: 라우팅 디폴트 4.7 → **4.8** (1M context 기본). Fable 5 명시 라우팅 신설. 공식 가이드 2종([Opus 4.8](https:/ - Network accessline 1915
- **Changes v2.8.0** (2026-06-10): - **[MAJOR] Claude 디폴트 = Opus 4.8 격상**: 라우팅 디폴트 4.7 → **4.8** (1M context 기본). Fable 5 명시 라우팅 신설. 공식 가이드 2종([Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-c - Network accessline 1916
- **[MAJOR] Claude 디폴트 = Opus 4.8 격상**: 라우팅 디폴트 4.7 → **4.8** (1M context 기본). Fable 5 명시 라우팅 신설. 공식 가이드 2종([Opus 4.8](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-4-8)·[Fable 5](https://pla - Network accessline 1918
- **[MEDIUM] 추천 모델 표·AskUserQuestion 옵션·모델별 필수 블록에 4.8/Fable 5 반영** (4.7/4.6 first-class 유지) - **Changes v2.6.0** (2026-04-30): - **[MAJOR] GPT-5.5 공식 outcome-first 가이드 반영**: [GPT-5.5 Prompt Guidance](https://developers.openai.com/api/doc - Network accessline 1919
- **Changes v2.6.0** (2026-04-30): - **[MAJOR] GPT-5.5 공식 outcome-first 가이드 반영**: [GPT-5.5 Prompt Guidance](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) 2026-04 발표. XML 12블록 stack → Markdown 6섹션 (Role / Per - Network accessline 1920
- **[MAJOR] GPT-5.5 공식 outcome-first 가이드 반영**: [GPT-5.5 Prompt Guidance](https://developers.openai.com/api/docs/guides/prompt-guidance?model=gpt-5.5) 2026-04 발표. XML 12블록 stack → Markdown 6섹션 (Role / Personality / Goal / Success Criteria / - Network accessline 1926
- **[LOW] Anti-Patterns 추가**: judgment 영역 ALWAYS/NEVER 회피, step sequence 강요 회피, retrieval 남용 회피, plain paragraph 디폴트 - **Changes v2.5.0** (2026-04-24): - **[MAJOR] Claude Opus 4.7 공식 프롬프팅 가이드 반영**: [platform.claude.com](https://platform.c - Network accessline 1927
- **Changes v2.5.0** (2026-04-24): - **[MAJOR] Claude Opus 4.7 공식 프롬프팅 가이드 반영**: [platform.claude.com](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Adaptive thinking](https://pl - Network accessline 1928
- **[MAJOR] Claude Opus 4.7 공식 프롬프팅 가이드 반영**: [platform.claude.com](https://platform.claude.com/docs/en/docs/build-with-claude/prompt-engineering/claude-4-best-practices) + [Adaptive thinking](https://platform.claude.com/docs/en/docs/build-
How the signals fit together
Facts from a static read of the files at the pinned commit. They describe what the code does, not why.
Some files could not be read; the facts below still hold.
Signal confidence: 172 facts
- Network access in prose of documentationGems-Prompt-Generator/02-role-model-rankings.md:22Low confidence
- Network access in prose of documentationGems-Prompt-Generator/05-workflow-generation-output.md:100Low confidence
- Network access in prose of documentationGems-Prompt-Generator/05-workflow-generation-output.md:102Low confidence
- Network access in prose of documentationGems-Prompt-Generator/07-references-final-reminder.md:43Low confidence
- Network access in prose of documentationGems-Prompt-Generator/07-references-final-reminder.md:44Low confidence
- Network access in prose of documentationGems-Prompt-Generator/07-references-final-reminder.md:45Low confidence
- Network access in prose of documentationGems-Prompt-Generator/07-references-final-reminder.md:46Low confidence
- Network access in prose of the SKILL.mdGems-Prompt-Generator/SKILL.md:9Low confidence
- Network access in prose of the SKILL.mdGems-Prompt-Generator/SKILL.md:10Low confidence
- Network access in prose of the SKILL.mdGems-Prompt-Generator/SKILL.md:11Low confidence
- Executable script in code of a file that runsscripts/prompt-output-gate.mjs:1High confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:9Low confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:10Low confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:11Low confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:127Low confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:128Low confidence
- Network access in prose of documentationclaude-4.7-prompt-strategies/references/full.md:129Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:20Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:21Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:22Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:52Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:174Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:232Low confidence
- Network access in prose of documentationskills/claude-fable-5-prompt-strategies.md:233Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:11Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:12Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:13Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:15Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:16Low confidence
- Network access in prose of documentationexpert-domain-priming/references/full.md:17Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:9Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:10Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:279Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:333Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:455Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:457Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:925Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:926Low confidence
- Network access in prose of documentationgpt-5.6-prompt-enhancement/references/full.md:927Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:12Low confidence
- Network access in prose of documentationskills/image-prompt-guide/references/full.md:12Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:13Low confidence
- Network access in prose of documentationskills/image-prompt-guide/references/full.md:13Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:14Low confidence
- Network access in prose of documentationskills/image-prompt-guide/references/full.md:14Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:15Low confidence
- Network access in prose of documentationskills/image-prompt-guide/references/full.md:15Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:16Low confidence
- Network access in prose of documentationskills/image-prompt-guide/references/full.md:16Low confidence
- Network access in prose of documentationimage-prompt-guide/references/full.md:21Low confidence
and 122 more
Works with
| Agent | Status | Evidence | Install to |
|---|---|---|---|
| Claude Code | Native | Manifest .claude-plugin/marketplace.json | .claude/skills/<skill-name>/ |
| Codex | Compatible | Standard SKILL.md instructions/Gems-Prompt-Generator/SKILL.md | .agents/skills/<skill-name>/ |
| Cursor | Compatible | Standard SKILL.md instructions/Gems-Prompt-Generator/SKILL.md | .cursor/skills/<skill-name>/ |
| Gemini CLI | Compatible | Standard SKILL.md instructions/Gems-Prompt-Generator/SKILL.md | .gemini/skills/<skill-name>/ |
| GitHub Copilot | Compatible | Standard SKILL.md instructions/Gems-Prompt-Generator/SKILL.md | .github/skills/<skill-name>/ |
| OpenCode | Compatible | Standard SKILL.md instructions/Gems-Prompt-Generator/SKILL.md | .opencode/skills/<skill-name>/ |
- Native The repository ships a manifest, folder or frontmatter field for this agent.
- Compatible A standard SKILL.md that this agent can read.
- Unknown No evidence either way.
Derived automatically from paths, manifests and frontmatter. Not tested on each agent.
