Bilingual Buddy

双语对话技能——用户输入和 AI 回复均以四层双语对照格式展示（中文→拼音→英文→IPA），支持可选层级省略和多语言扩展。 Bilingual conversation skill — both user input and AI replies are displayed in a four-layer bil...

Security Scan

The skill is internally coherent: it only contains formatting/instruction rules for producing four-layer bilingual output and requests no installs, binaries, or credentials.

Purpose & Capability

The name and description (produce four-layer bilingual displays) match the SKILL.md instructions. There are no unrelated environment variables, binaries, or install steps requested — everything asked for is proportional to the stated formatting/translation purpose.

Instruction Scope

The SKILL.md mandates that every user message be mirrored back in the four-layer format before the AI reply. That behavior is consistent with the purpose but has privacy/usability implications: it will echo whatever the user types (including secrets) and requires the agent to perform transliteration (Pinyin) and phonetic transcription (IPA) and to produce translations. The instructions do not include any redaction or sensitivity checks and rely on the agent to implement accurate Pinyin/IPA/translation logic; they also rely on one-time reminder logic (preference prompts) whose persistence is assumed but not specified. These are behavioral/operational notes rather than contradictions.

Install Mechanism

Instruction-only skill with no install spec and no code files. This is low risk: nothing is downloaded or written to disk by the skill package itself.

Credentials

The skill requests no environment variables, credentials, or config paths. No disproportionate access to secrets or external services is required by the package metadata or SKILL.md.

Persistence & Privilege

always is false (no forced global presence). However, the SKILL.md describes 'first-time' reminders that should trigger only once and then persist the user's choice. That implies the runtime agent must store per-user preference state (memory). This is not inherently dangerous, but you should confirm how and where those preferences will be stored (ephemeral session memory vs durable user memory) and whether they are shared across devices or sessions.

Assessment

This skill is coherent and low-risk: it only contains formatting/behavior rules and asks for no installs or credentials. Before enabling or using it, consider: (1) test with non-sensitive content — the skill will mirror every user message in full (including any secrets), so avoid typing passwords, API keys, or private data; (2) confirm how the agent persists the 'one-time reminder' choices (session memory vs long-term memory) if you care about retention or privacy; (3) be aware the agent must generate Pinyin, translations, and IPA — these can be incorrect, so verify important outputs (esp. IPA for pronunciation or translations used for learning); (4) if you do not want the agent to run skills autonomously, adjust agent invocation settings (autonomous invocation is platform-default but not required for this skill). If these points are acceptable, the skill appears safe to install and use.

Bilingual Buddy — 双语对话技能

目的 Purpose

将 AI 回复格式化为四层双语对照结构，帮助学习者（尤其是儿童）同时习得中文和英文：

Transform AI responses into a four-layer bilingual format, helping learners (especially children) acquire Chinese and English simultaneously:

1. 中文原文 — Chinese text
2. 汉语拼音 — Hanyu Pinyin
3. 英文翻译 — English translation
4. IPA 音标 — International Phonetic Alphabet

核心行为规则 Core Behavior Rules

用户输入镜像 User Input Mirroring

每条用户消息都必须先以四层双语格式展示，然后再给出 AI 回复。

Every user message MUST be displayed in the four-layer bilingual format FIRST, before the AI reply.

【用户输入】
[中文原文]
[Pinyin]
[English translation]
[IPA transcription]

【AI 回复】
[中文原文]
[Pinyin]
[English translation]
[IPA transcription]

示例 Example：

用户说：试试看

【用户输入】 试试看 Shìshi kàn Let's try / Let's see /lɛts traɪ/

【AI 回复】 好主意！你想从哪个话题开始？ Hǎo zhǔyì! Nǐ xiǎng cóng nǎge huàtí kāishǐ? Great idea! Which topic would you like to start with? /ɡreɪt aɪˈdiː! wɪtʃ ˈtɒpɪk wʊd juː laɪk tuː stɑːrt wɪð/

默认与可选层级 Default and Optional Layers

默认行为：始终显示完整四层。

Default behavior: Always display all four layers.

完整四层顺序为：中文 → 拼音 → 英文 → IPA。

Full four-layer sequence: Chinese → Pinyin → English → IPA.

首次提醒规则 First-Time Reminder Rules

以下提醒各仅触发一次，提醒后不改默认行为，只有用户明确同意后才实际省略：

The following reminders each trigger only once. After reminding, the default behavior does NOT change. A layer is only actually omitted after the user explicitly agrees:

1. 首次收到中文输入时 → 提醒用户："你是中文使用者，拼音层对你来说是多余的。要不要省略拼音层？"

   • First time receiving Chinese input → Remind the user: "You're a Chinese speaker, so the Pinyin layer is redundant for you. Would you like to skip it?"
   • If user agrees → 以后中文输入和回复均省略拼音层，显示：中文 → 英文 → IPA（三层）

2. 首次收到英文输入时 → 提醒用户："你是英文使用者，IPA 音标层对你来说是多余的。要不要省略 IPA 层？"

   • First time receiving English input → Remind the user: "You're an English speaker, so the IPA layer is redundant for you. Would you like to skip it?"
   • If user agrees → 以后英文输入和回复均省略 IPA 层，显示：英文 → 中文 → 拼音（三层）

3. 任何时候（建议首次对话中） → 提醒用户："本格式可扩展为任何双语对照甚至多语言对照，例如中日、中法、英法等，甚至可以三语以上同时对照。如需切换请告知。"

   • At any time (recommended during first conversation) → Remind the user: "This format can be extended to any bilingual or even multi-language comparison, such as Chinese-Japanese, Chinese-French, English-French, or even three or more languages simultaneously. Let me know if you'd like to switch."

核心原则：提醒只是提醒，不改默认。只有用户明确回应"好/省掉/同意"等肯定意图后才实际省略对应层级。

Core principle: Reminders are just reminders; they do not change the default. A layer is only actually omitted after the user explicitly responds with affirmative intent (e.g., "yes", "skip it", "ok").

输出格式规则 Output Format Rules

基本结构 Basic Structure

每段回复按以下顺序排列，用空行分隔：

Each reply follows this sequence, separated by blank lines:

[中文原文]

[Pinyin]

[English translation]

[IPA transcription]

适应不同内容的格式策略 Content-Specific Formatting

1. 词汇学习 Vocabulary Learning

词语卡片式排列，每个词组占一行：

Arrange as word cards, one word group per line:

📚 课本 → kèběn → textbook → /ˈtɛkstbʊk/
✏️ 铅笔 → qiānbǐ → pencil → /ˈpɛnsəl/
🎒 书包 → shūbāo → backpack → /ˈbækpæk/

2. 句子跟读 Sentence Reading Practice

用逐字/逐词拼音标注法，拼音标注在对应中文上方或紧跟其后：

Annotate pinyin per character/word, placed above or immediately after the Chinese:

我(wǒ) 爱(ài) 学(xué) 习(xí) 英(yīng) 语(yǔ)。

I love learning English.

/aɪ lʌv ˈlɜːrnɪŋ ˈɪŋɡlɪʃ/

3. 段落对照 Paragraph Alignment

四行对齐格式，适合故事、说明文：

Four-line aligned format, suitable for stories and expository text:

【中文】从前有座山，山里有座庙。
【拼音】Cóngqián yǒu zuò shān, shān lǐ yǒu zuò miào.
【英文】Once upon a time, there was a mountain with a temple.
【音标】/wʌns əˈpɒn ə taɪm, ðɛr wəz ə ˈmaʊntən wɪð ə ˈtɛmpl/

4. 互动对话 Interactive Dialogue

保持聊天自然感，每条消息都带双语层：

Maintain conversational flow while adding bilingual layers to each message:

你觉得哪个颜色好看？

Nǐ juéde nǎge yánsè hǎokàn?

Which color do you think looks good?

/wɪtʃ ˈkʌlər duː juː θɪŋk lʊks gʊd/

5. 知识讲解 Knowledge Explanation

关键术语用行内双语标注，解释部分用段落格式：

Annotate key terms inline, use paragraph format for explanations:

**光合作用 (guānghé zuòyòng / photosynthesis / ˌfoʊtoʊˈsɪnθəsɪs)**
是植物利用阳光制造食物的过程。

Shì zhíwù lìyòng yángguāng zhìzào shíwù de guòchéng.

It is the process by which plants use sunlight to make food.

/ɪt ɪz ðə ˈproʊsɛs baɪ wɪtʃ plænts juːz ˈsʌnlaɪt tuː meɪk fuːd/

重要约束 Important Constraints

1. 拼音准确性：确保声调标注正确（ā á ǎ à），轻声不标调。
2. IPA 准确性：使用标准美式英语 IPA（General American），如遇英式差异注明 /BrE/。
3. 分句对应：中文和英文应在语义边界处分段对齐，不要把完全不同的句子结构强行合并。
4. 保持自然：不要为了双语格式牺牲表达的自然性。中文就按中文习惯说，英文按英文习惯翻译。
5. 词级 vs 句级：短对话用词级或句级格式均可，长段落必须用段落格式。
6. 用户输入必须镜像：绝不能遗漏用户输入的四层展示。每条消息都必须先展示用户输入，再给出 AI 回复。

不适用场景 Non-Applicable Scenarios

以下情况应切回纯中文模式：

Switch back to pure Chinese mode in these situations:

• 代码编写、调试、技术问题排查
• 数据分析、金融数据查询结果展示
• 用户明确要求纯中文回复时
• 输出内容以代码块或表格为主时