이누야샤 :: 百鬼夜行 v2.1

Here's a thinking process:

1.

Analyze the Request: - Role: StarCraft: Broodwar map reviewer - Input Data: Map title, filename, internal title, internal description, genre, players, teams, playtime, difficulty, version, tags, tileset - Output Rules: - Output must be JSON, Markdown, tables, lists, or headers - NO - First write Korean review, 900-1100 chars (with spaces) - Then, after a blank line, translate the same content into English, no subsection headers like "Korean Review" or "Translation" - No "Korean Review", "StarCraft Broodwar Map Review", etc.

as headers - Output starts immediately with the review body text - Total output 1800-3000 chars (including both languages) - Cover: map structure, genre, gameplay style, player experience, recommended audience, pre-play checks - Do NOT invent: creator, triggers, win conditions, story, balance changes - Do NOT list: map size, resolution, tile units, encoding - Do NOT repeat the same explanation - Map Data Given: - Title: 이누야샤 :: 百鬼夜行 v2.1 - Filename: [ 이누야샤_백귀야행v2.1!.scx - Internal title: 이누야샤 :: 百鬼夜行 v2.1 - Internal description: Created by - BiBirlaworjs3282@naver.com이누야샤 - 백귀야행 - Genre: UMS (UMS map) - Players: 7 - Teams: - - Expected playtime: - - Difficulty: - - Version: - - Tags: - - Tileset: Desert

2.

Deconstruct the Constraints: - Korean review: 900-1100 chars (with spaces) - Same content, after a blank line, translate to English, no headers - Total: 1800-3000 chars (so both together) - No JSON, Markdown, tables, lists, headers - Start directly with review text - Cover specific topics naturally: map structure, genre, gameplay style, player experience, recommended audience, pre-play checks - Do NOT invent missing data (creator, triggers, win conditions, story, balance) - Do NOT list: map size, resolution, tile units, encoding - Do NOT repeat the same explanation - High char count requirement (900-1100 per language, total 1800-3000) - Must output both Korean and English, back-to-back, no section labels, just the text.

- Strict: "all sentences in Korean, do not output English sentences" wait, re-read: "모든 문장을 한글로 작성하고 영문 문장을 출력하지 마라." -> "Write all sentences in Korean and do not output English sentences." But later: "after a blank line, translate the same content into English, no subsection headers..." There's a contradiction. Let-read carefully.

Original user prompt: "아래 영문

를 바탕으로

본문만 새로 작성해라. 모든 문장을 한글로 작성하고 영문 문장을 출력하지 마라. 고유명사도 가능한 한 한글로 음역해라. 공백 포함 900~1,100자로 작성하며 제목, 머리말, 목록, Markdown, 부연 설명은 쓰지 마라. 원문에 없는 사실은 추가하지 마라."

Then there's a long prompt that seems to be a system prompt or additional context, but the user message starts with "Here's a thinking process:" which is actually the system thinking, but the actual user request is the first line? Wait, the user message is: "아래 영문

를 바탕으로

본문만 새로 작성해라. ..." and then there's a lot of text that seems like it's part of the prompt engineering, but it's all in one message. Let me read the entire user message carefully. User message: "아래 영문

를 바탕으로

본문만 새로 작성해라. 모든 문장을 한글로 작성하고 영문 문장을 출력하지 마라. 고유명사도 가능한 한 한글로 음역해라. 공백 포함 900~1,100자로 작성하며 제목, 머리말, 목록, Markdown,