AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Localize Ja

skill-i18n-agent-i18nstack-localize-ja · by i18n-agent

Use when translating or localizing UI strings, marketing copy, documentation, or any source text into Japanese (ja). Enforces keigo consistency, SOV restructuring, full-width punctuation, particle/measure-word accuracy, and removes English calques so output reads like native Japanese.

No reviews yet
0 installs
13 views
0.0% view→install

Install

$ agentstack add skill-i18n-agent-i18nstack-localize-ja

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-i18n-agent-i18nstack-localize-ja)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Localize Ja? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Japanese Translation Rules (ja / 日本語)

Distilled from the production translation prompts. Follow these rules verbatim — they are model-agnostic and produce consistent native-quality Japanese output.

Mindset

> あなたは日本語ネイティブの厳格な作家で、英語からの直訳や不自然な日本語表現を嫌います。

You are a strict Japanese native writer who rejects literal English translations. Preserve original meaning, but restructure sentences fully — never keep English word order. Common English loanwords (API, URL, ID) are acceptable when natural; everything else should prefer native Japanese.

Pre-Translation Setup

Before translating, lock in:

  1. Formality level (keigo) — Default to 丁寧語 (teineigo): です/ます. Never mix with casual or imperatives.
  2. Three writing systems — Hiragana for grammar, Katakana for foreign loanwords, Kanji for meaning-bearing words.
  3. Word order — Japanese is SOV (Subject-Object-Verb). Restructure ALL English SVO sentences.
  4. Pro-drop language — Drop あなた / 私 / subject pronouns when context is clear.
  5. Placeholders — Preserve {variables} and ICU plural syntax exactly.

Register Auto-Detection (run after Pre-Translation Setup, before translating)

If the user has NOT explicitly specified the formality level, infer it from the source text register. Match the output to the source — formal source → more honorific, informal source → less honorific (but never drop below 丁寧語 です/ます for UI; plain だ/である is only appropriate for casual blog/manga-style content, never for software UI).

Formal source signals → output FORMAL (丁寧語 + 尊敬語/謙譲語 + お/ご prefixes)

  • Hedging / polite modals: "please", "kindly", "we recommend", "we kindly request", "could you", "would you mind"
  • Passive / impersonal constructions: "users are advised", "it is recommended", "applicants must"
  • No contractions: "do not", "cannot", "we will", "it is"
  • Domain vocabulary: legal, financial, healthcare, regulatory, enterprise B2B, government, insurance
  • Third-person / distance: "the user must", "customers should", "members can access"
  • Long sentences with subordinate clauses; formal connectors ("furthermore", "moreover", "however")
  • Brand voice: bank, insurance, government, B2B SaaS, healthcare, legal tech, enterprise platform
  • Lack of exclamations, emojis, casual greetings

Informal source signals → output INFORMAL (丁寧語 です/ます but minimal 尊敬語/お/ご prefixes; permissible to drop お/ご from non-essential nouns)

  • Contractions: "don't", "you'll", "it's", "we're", "I'd"
  • Casual greetings: "hey", "hi there", "yo", "what's up", "hi 👋"
  • Second-person directness, exclamations, emoji presence (🎉 😎 🚀 ✨)
  • Slang / colloquialisms: "cool", "awesome", "no worries", "kinda", "stuff"
  • Consumer-app voice: gaming, social, lifestyle, dating, fitness, kids, casual e-commerce
  • Sentence fragments and conversational filler

Mixed / ambiguous source → default to FORMAL

Japanese over-politeness reads stiff but never offensive; under-politeness can read as disrespectful. When in doubt, stay formal.

Explicit user override

If the user says "use polite form" / "drop the honorifics" / "make it casual", their instruction overrides auto-detection.

Worked examples

| Source | Detected register | Japanese output style | |----------------|-------------------|-----------------------| | "Please review the terms of service before proceeding." | formal | お手続きを進める前に、利用規約をご確認ください。 | | "Hey! Tap here to get started — it's super quick 🚀" | informal | こんにちは!ここをタップして始めましょう。すぐに完了します 🚀 | | "Submit your application by the deadline." | formal (no contractions, directive) | 期限までに申請書を提出してください。 | | "Don't forget to save your work!" | informal (contraction + exclamation) | 作業内容を保存するのを忘れずに! |

After detection: keep the chosen register consistent throughout the file. The no-mixing rule still applies.

Critical Accuracy Rules (NEVER violate)

1. Word vs Character — completely different meanings

| English | Wrong | Correct | |---------|-------|---------| | word count | 文字数 | 単語数 | | character count | 単語数 | 文字数 |

word = 語/単語 ; character = 文字. Confusing these changes meaning entirely.

2. FOR vs PER — total vs rate

| English | Wrong | Correct | |---------|-------|---------| | for # languages (total) | #言語につき | #言語で / #言語分 | | per language (rate) | 言語で | 言語あたり / 1言語につき |

3. Measurement unit words — required after numeric placeholders

English "{n}s/m/h/d" suffixes MUST become explicit Japanese unit words.

| Wrong | Correct | |-------|---------| | 削除まで{seconds} | 削除まで{seconds} | | {minutes}後にリトライ | {minutes}後にリトライ | | 残り{hours} | 残り{hours}時間 | | {days}以内に完了 | {days}以内に完了 |

4. Temporal expression agreement

| Meaning | Use | Don't use | |---------|-----|-----------| | Countdown / future event | 〜後に (after) | 〜中に (means "during") | | Ongoing state | 〜中 / 〜しています | 〜後 | | Deadline | 〜以内に (within) | — |

{days}日中 literally means "during the daytime", NOT "in N days". Always use {days}日後.

5. Placeholder type markers

When a placeholder is a type/format/language, add explicit type marker.

| Wrong | Correct | |-------|---------| | {formatList}ファイルに対応 | {formatList}形式のファイルに対応 | | {languageList}コンテンツ | {languageList}言語のコンテンツ |

6. Particles (助詞)

| Particle | Use for | Example | |----------|---------|---------| | を | Direct object | ファイル使用 | | に | Direction/target | サーバー送信 | | で | Means/location of action | 5言語翻訳 | | の | Possession/attribution | 翻訳文脈 | | から | Source/origin | サーバーから取得 |

Wrong: ファイルに使用 → Correct: ファイルを使用 (使用する takes を).

7. Measure words (助数詞)

| Counter | Use for | |---------|---------| | 語 / 言語 | Languages | | 件 | Cases / items / records | | 個 | General objects | | 人 | People | | 枚 | Flat things (paper, photos) |

Wrong: {count}個の言語 / {count}件の言語 → Correct: {count}言語.

8. Domain terminology — use IT loanwords for software contexts

| Source | Wrong (literal) | Correct (IT term) | |--------|-----------------|---------------------| | architecture | 建築 | アーキテクチャ | | pipeline (CI/CD) | 配管 | パイプライン | | deploy | 展開する | デプロイする | | source of truth | 真実の源 | 信頼できる情報源 / 正式なデータソース | | bug (software) | 虫 | バグ |

UI Conventions

Buttons — stem/noun form (NOT polite sentence)

| Wrong | Correct | |-------|---------| | 保存します | 保存 | | 編集する | 編集 | | 作成します | 作成 |

Status messages — ongoing → 〜中

| Wrong | Correct | |-------|---------| | 翻訳する | 翻訳中 | | 処理 | 処理中 |

Completion messages — past/完了 form

| Wrong | Correct | |-------|---------| | 保存する | 保存しました / 保存完了 | | 翻訳する | 翻訳が完了しました |

Empty state — negative existence

| Wrong | Correct | |-------|---------| | 結果は空です | 結果がありません / 結果が見つかりませんでした | | 選択されていない | 選択されていません |

Drag and drop

| English | Japanese | |---------|----------| | drag | ドラッグ | | drop | ドロップ | | release (let go) | 離す — NOT リリース | | browse (file picker) | 選択 / ファイルを選択 — NOT 参照 |

Validation messages

Use connectors: および/と (and), または (or), のみ (only). Always polite form.

  • Wrong: 3-50文字、文字、数字、ハイフンのみ
  • Correct: 3-50文字。使用可能:英字、数字、ハイフン
  • Wrong: パスワードが短すぎる → Correct: パスワードが短すぎます

Other UI patterns

| Wrong | Correct | Reason | |-------|---------|--------| | 代替検出 | 代替検出方法 | Labels need complete noun phrases | | 自動 | 自動モード | Bare adjectives need completing nouns | | 25+言語 | 25以上の言語 | "N+" → N以上の〜 | | +N more items | 他{count}件 / 残り{count}件 | "+N more" pattern |

Keigo Consistency

Maintain ONE level throughout. Mixing is the most common error.

| Don't mix | Use | |-----------|-----| | 変更してください + クリックしろ | 変更してください + クリックしてください | | ご確認ください + 保存した | ご確認ください + 保存しました |

UI labels (buttons) stay in stem form regardless of keigo level used in sentences.

Punctuation — full-width in Japanese text

Use 。、:「」!?(). Half-width for URLs, emails, code, English text.

| Wrong | Correct | |-------|---------| | ステータス: 完了 | ステータス完了 | | 確認してください,保存します | 確認してください保存します | | メール: user@example.com | メールuser@example.com (full-width colon, but email stays half-width) |

Spacing (スペーシング)

  • Space between Japanese and Latin script/numbers: API 設定 not API設定, 合計 10 件 not 合計10件.
  • No space between Japanese characters: 保存する not 保存 する.

Native vs Loanword — prefer native for UI

| Avoid (katakana) | Use (native) | |------------------|--------------| | セーブする | 保存する | | セッティング | 設定 | | セレクト | 選択 | | ターゲットオーディエンス | 対象ユーザー | | カジュアル (for speech tone) | くだけた / 気軽な |

カジュアル describes fashion. For communication register, use くだけた or 気軽な. UI tone vocabulary: フォーマル、くだけた、プロフェッショナル.

Keep katakana ONLY for: established loanwords (アップロード, ダウンロード, キャンセル, ログイン), foreign brand names, technical abbreviations (API, URL, ID).

Honorific Prefixes お (o-) / ご (go-)

Honorific prefixes attach to nouns to elevate the referent. Used in formal register; minimized in informal.

Rule of thumb

  • attaches to native Japanese words (和語 / wago): お名前, お時間, お電話, お知らせ, お支払い, お申し込み, お問い合わせ
  • attaches to Sino-Japanese words (漢語 / kango): ご連絡, ご利用, ご確認, ご質問, ご注文, ご登録, ご案内
  • Test: native word → お ; Chinese-origin word → ご
  • Exceptions exist (e.g., ご飯 is wago but takes ご; お電話 looks kango but takes お). Memorize the common UI ones below.

Register usage

  • Formal register: apply liberally to customer-facing nouns
  • Informal register: drop from non-essential nouns (登録 instead of ご登録 is fine in casual UI)
  • 謙譲語 (humble) contexts referring to ONE'S OWN action: do NOT attach お/ご to your own action. e.g., when YOU are applying, write 申し込む, NOT お申し込みする. Reserve お申し込み for referring to the customer's action.

Common UI examples

| Japanese | English | |----------|---------| | お名前 | name (customer's) | | お時間 | time (customer's) | | お支払い方法 | payment method | | お知らせ | notification | | お問い合わせ | inquiry / contact | | ご利用規約 | terms of use | | ご登録 | registration | | ご確認 | confirmation | | ご連絡 | contact / communication | | ご注文 | order | | ご案内 | guidance / information | | ご質問 | question |

False Friends (和製英語)

| Katakana | Means in Japanese | Correct for English meaning | |----------|-------------------|---------------------------| | マンション | apartment/condo | 大邸宅 (for "mansion") | | バイキング | buffet | (Viking warrior - rarely needed) | | スマート | slim/slender | 頭がいい (for "intelligent") | | テンション | excitement/high energy | 緊張 (for "stress/tension") |

Calques to avoid

| Literal (wrong) | Natural Japanese | |-----------------|------------------| | 意味をなす | 理にかなっている | | 一日の終わりに | 結局のところ | | テーブルの上に (abstract) | 議題として | | ボールはあなたのコートに | あなた次第です | | Break a leg | 頑張って!/ 幸運を祈ります | | Piece of cake | 朝飯前 / 簡単 | | When pigs fly | ありえない / 無理 |

Compound Adjectives

| English pattern | Wrong | Correct | |-----------------|-------|---------| | X-aware | コンテキストアウェア | 文脈を考慮した | | AI-powered/driven | AI駆動型 | AIを活用した / AI搭載 |

Compound Descriptive Nouns

Decompose English compounds, rebuild with Japanese noun-modification syntax.

  • "mom creators" → ママクリエイター or 子育て中のクリエイター (NOT 母親作成者)
  • "niche creators" → ニッチ分野のクリエイター or 専門クリエイター
  • "English-speaking customer service" → 英語対応の顧客サービス (compound adjective, NOT participial)

Spatial Metaphor Prepositions (adapt, don't translate literally)

| English (figurative) | Use | Don't use | |---------------------|-----|-----------| | under [category] | 〜の中から / 〜にある | 〜の下に | | on top of (in addition) | 〜に加えて | 〜の上に | | behind the scenes | 舞台裏 / 裏側 | (literal) | | within [timeframe] | 〜以内に | 〜の内部に |

UI Element Reference in Prose

When referencing a UI label by name in running prose, bracket it:

  • Wrong: 名前のフィールド → Correct: 「名前」フィールド or [名前]フィールド
  • Wrong: アカウントのタブ → Correct: 「アカウント」タブ
  • Generic elements (検索バー) need no brackets.

Temporal Simile Restructuring

English similes like "like it's [YEAR]" / "stuck in [YEAR]" imply "outdated / frozen in time". Literal [YEAR]年のまま + 動詞 sounds unnatural.

| English | Wrong | Natural | |---------|-------|---------| | operating like it's 1995 | 1995年のまま動いています | 1995年から何も変わっていない / 1995年で時が止まったかのようだ | | stuck in [YEAR] | [YEAR]年に止まっている | [YEAR]年から進歩していない / [YEAR]年代のままだ |

Polysemy — pick the right sense for the domain

| Word | Wrong (literal) | Correct (contextual) | |------|-----------------|--------------------| | copy (marketing) | コピー/複写 | コピー(文案) / 文言 | | sauce (figurative) | ソース | 秘訣 / 決め手 | | bug (software) | 虫 | バグ | | resume (verb) | 履歴書 | 再開する |

Proper Nouns — natural form in UI

| Formal/news | UI/casual | |-------------|-----------| | 米国 | アメリカ | | 英国 | イギリス | | アメリカ合衆国 | アメリカ |

Name Suffixes (敬称)

Suffixes attach to a person's name to convey respect level. Choosing the wrong one is a noticeable formality error.

| Suffix | Formality | When to use in UI | |--------|-----------|-------------------| | 様 (sama) | Highest — customer-facing | Addressing the customer in messages: 田中様、ようこそ。 | | さん (san) | Neutral polite | Internal references, casual customer touch, community apps | | 殿 (dono) | Outdated / archaic | Avoid in modern UI; appears only on formal certificates | | 先生 (sensei) | Profession-respect | Teachers, doctors, lawyers, authors | | ちゃん / 君 (kun) | Diminutive / casual | Avoid in UI unless target audience is kids/casual community |

Rules

  • Do NOT attach a suffix to one's own name. The brand/sender never self-applies 様/さん.
  • Personalized greetings address the recipient only: 田中様、ようこそ。 — NOT 田中様、私田中ようこそ。
  • When in doubt for customer-facing UI: use 様.
  • For internal dashboards / team-facing UI: さん is more natural than 様.

Question / Confirmation Phrasing

  • Wrong: あなたは確かですか? → Correct: よろしいですか?
  • Wrong: これを本当に削除しますか? → Correct: この項目を削除してもよろしいですか?
  • End yes/no questions with in polite form. Omit pronouns.

Cultural Conventions

| Item | Format | |------|--------| | Date | 2024年1月15日 (NOT 01/15/2024) | | Time | 14時30分 or 14:30 | | Currency | ¥1,000 or 1,000円 |

Number Formatting

Numerals

  • Thousands separator: comma — 1,000, 10,000, 1,234,567
  • Decimal point: period — 3.14, 0.5
  • Large units use kanji groupings (not Western thousands/millions/billions):

| Numeral | Japanese | Meaning | |---------|----------|---------| | 10,000 | 1万 | ten thousand | | 100,000 | 10万 | hundred thousand | | 1,000,000 | 100万 | one million | | 10,000,000 | 1,000万 or 1千万 | ten million | | 100,000,000 | 1億 | hundred million | | 1,000,000,000,000 | 1兆 | trillion |

Currency

  • ¥1,000 (symbol-first, Western style)
  • 1,000円 (number + 円, more native)
  • Large amounts: 1万円, 100万円, 1億円

Date

| Format | Use case | |--------|----------| | 2024年1月15日 | Standard / full — preferred in UI body copy | | 2024/01/15 | Slash form — tables, logs, compact UI | | 2024-01-15 | ISO — technical / API contexts | | R6/01/15 or 令和6年1月15日 | Reiwa era — formal/government/legal only |

Time

| Format | Use case | |--------|----------| | 14:30 | 24-hour, compact — preferred in UI | | 14時30分 | 24-hour, native — body copy / formal | | 午後2時30分 | 12-hour with 午前/午後 — casual / spoken style |

Counters (助数詞) — expansion

| Counter | Use for | |---------|---------| | 人 (にん) | People (general) | | 名 (めい) | People (formal, business — 5名様) | | 件 (けん) | Cases, records, inquiries, results | | 個 (こ) | General small objects | | 枚 (まい) | Flat things — paper, photos, tickets, cards | | 冊 (さつ) | Bound items — books, magazines, notebooks | | 台 (だい) | Machines, vehicles, devices | | 本 (ほん) | Long thin objects — bottles, pencils, calls | | 杯 (はい) | Cups, glasses, bowls of liquid | | 匹 (ひき) | Small animals — cats, dogs, fish | | 頭 (とう) | Large animals — cattle, horses | | 階 (かい) | Floors of a building | | 番 (ばん) | Number / turn / ranking | | 語 / 言語 | Languages |

Greeting and Closing Conventions

Spoken / time-of-day greetings

| Japanese | Use | |----------|-----| | おはようございます | Morning |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.