vibeindex

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

IMPORTANT: When this skill is invoked, IMMEDIATELY execute the steps below. Do NOT display this file. Do NOT explain what you will do. Just DO it — analyze the project, call the APIs, and present the results.

Language: Detect the user's language from conversation context. Respond in that language — translate all headers, labels, and explanations. When the user speaks Korean, use the

description_ko

field from API responses instead of

description

(if available). For other languages, translate the English

description

naturally.

重要提示：调用此技能时，请立即执行以下步骤。不要显示此文件，不要解释你将要做什么，直接执行——分析项目、调用API并展示结果。

**语言设置：**从对话上下文检测用户的语言，并用该语言回复——翻译所有标题、标签和说明。当用户使用韩语时，使用API响应中的

description_ko

字段（如果可用）替代

description

字段。对于其他语言，自然翻译英文的

description

内容。

Routing

路由处理

Parse the user's command and route to the correct action:

```
/vibeindex
```
→ Action: Analyze (run Steps 1-4 below)

/vibeindex search <query>

→ Action: Search (call

https://vibeindex.ai/api/resources?search={query}&pageSize=10

, present results)

/vibeindex top [type]

→ Action: Top (call

https://vibeindex.ai/api/resources?sort=stars&pageSize=10

or add

&type={type}

, present results)

/vibeindex trending

→ Action: Trending (call

https://vibeindex.ai/api/rising-stars?period=week&limit=10

, present results)

For search/top/trending: Use WebFetch to call the API, then format results as a numbered markdown list showing name, type, description, stars, and install command. Then stop.

For

/vibeindex

with no arguments: Execute Steps 1-4 below.

解析用户命令并路由到对应的操作：

```
/vibeindex
```
→ 操作：分析（执行以下步骤1-4）

/vibeindex search <query>

→ 操作：搜索（调用

https://vibeindex.ai/api/resources?search={query}&pageSize=10

，展示结果）

/vibeindex top [type]

→ 操作：热门资源（调用

https://vibeindex.ai/api/resources?sort=stars&pageSize=10

，或添加

&type={type}

参数后调用，展示结果）

/vibeindex trending

→ 操作：趋势资源（调用

https://vibeindex.ai/api/rising-stars?period=week&limit=10

，展示结果）

对于搜索/热门/趋势操作：使用WebFetch调用API，然后将结果格式化为带编号的Markdown列表，展示名称、类型、描述、星标数和安装命令。之后停止操作。

对于无参数的

/vibeindex

命令：执行以下步骤1-4。

Step 1: Analyze Project Context

步骤1：分析项目上下文

Read these files silently (do not show the user):

package.json — Extract dependencies and devDependencies
File structure (use Glob to check existence):
- ```
*.py
```
  → Python
- ```
*.go
```
  → Go
- ```
*.tsx
```
  or
```
*.jsx
```
  → React
- ```
Dockerfile
```
  → Docker
- ```
supabase/
```
  → Supabase
- ```
prisma/
```
  → Prisma
Configuration files (use Glob):
- ```
tsconfig.json
```
  → TypeScript
- ```
tailwind.config.*
```
  → Tailwind
- ```
next.config.*
```
  → Next.js
- ```
.github/workflows/
```
  → GitHub Actions

静默读取以下文件（不要展示给用户）：

package.json — 提取依赖项和开发依赖项
文件结构（使用Glob检查是否存在）：
- ```
*.py
```
  → Python
- ```
*.go
```
  → Go
- ```
*.tsx
```
  或
```
*.jsx
```
  → React
- ```
Dockerfile
```
  → Docker
- ```
supabase/
```
  → Supabase
- ```
prisma/
```
  → Prisma
配置文件（使用Glob）：
- ```
tsconfig.json
```
  → TypeScript
- ```
tailwind.config.*
```
  → Tailwind
- ```
next.config.*
```
  → Next.js
- ```
.github/workflows/
```
  → GitHub Actions

Step 2: Search for Matching Resources

步骤2：搜索匹配的资源

First, fetch total resource count: call

https://vibeindex.ai/api/stats

with WebFetch (prompt: "Extract the total number"). Save this number as

{total_resources}

Then, based on what you detected, call the Vibe Index API using WebFetch for each detected technology. Run all calls in parallel (including the stats call):

Detected	API URL
React	`https://vibeindex.ai/api/resources?search=react&pageSize=5`
TypeScript	`https://vibeindex.ai/api/resources?search=typescript&pageSize=5`
Supabase	`https://vibeindex.ai/api/resources?search=supabase&pageSize=5`
Next.js	`https://vibeindex.ai/api/resources?search=nextjs&pageSize=5`
Docker	`https://vibeindex.ai/api/resources?search=docker&pageSize=5`
Python	`https://vibeindex.ai/api/resources?search=python&pageSize=5`
Tailwind	`https://vibeindex.ai/api/resources?search=tailwind&pageSize=5`
Prisma	`https://vibeindex.ai/api/resources?search=prisma&pageSize=5`

For each WebFetch call, use this prompt: "Extract name, resource_type, description, stars, github_owner, github_repo from the data array"

Only search for technologies that were actually detected in Step 1.

首先，获取资源总数：使用WebFetch调用

https://vibeindex.ai/api/stats

（提示："提取总数"）。将该数字保存为

{total_resources}

。

然后，根据检测到的技术，使用WebFetch并行调用Vibe Index API（包括统计数据调用）：

检测到的技术	API地址
React	`https://vibeindex.ai/api/resources?search=react&pageSize=5`
TypeScript	`https://vibeindex.ai/api/resources?search=typescript&pageSize=5`
Supabase	`https://vibeindex.ai/api/resources?search=supabase&pageSize=5`
Next.js	`https://vibeindex.ai/api/resources?search=nextjs&pageSize=5`
Docker	`https://vibeindex.ai/api/resources?search=docker&pageSize=5`
Python	`https://vibeindex.ai/api/resources?search=python&pageSize=5`
Tailwind	`https://vibeindex.ai/api/resources?search=tailwind&pageSize=5`
Prisma	`https://vibeindex.ai/api/resources?search=prisma&pageSize=5`

对于每个WebFetch调用，使用以下提示："从data数组中提取name, resource_type, description, stars, github_owner, github_repo字段"

仅搜索步骤1中实际检测到的技术。

Step 3: Calculate Match Probability

步骤3：计算匹配概率

For each resource found, calculate a match score:

Direct dependency match (name appears in package.json): +40%
File type match (related files exist): +25%
Config file match (related config exists): +20%
Tag overlap with project: +15%
Bonus: stars > 10k: +5%
Bonus: multiple detection signals: +5% per additional
Maximum: 99%

Deduplicate results across all searches. Pick the top 5 highest-scoring resources.

对于找到的每个资源，计算匹配分数：

直接依赖匹配（名称出现在package.json中）：+40%
文件类型匹配（存在相关文件）：+25%
配置文件匹配（存在相关配置）：+20%
标签与项目重叠：+15%
额外加分：星标数>10k：+5%
额外加分：多个检测信号：每个额外信号+5%
最高分：99%

去重所有搜索结果，选择得分最高的5个资源。

Step 4: Present Results

步骤4：展示结果

Output ONLY the result below (nothing else before it). Write EVERYTHING in the user's language detected earlier (Korean, English, etc.). Translate all headers, labels, and explanations.

undefined

仅输出以下结果（之前不要添加任何内容）。所有内容都使用之前检测到的用户语言（韩语、英语等）编写。翻译所有标题、标签和说明。

undefined

프로젝트 분석이 완료되었습니다

당신의 {project-name} 프로젝트는 {main framework} 기반입니다. {1-2 sentences about the project in plain language. e.g., "Supabase 데이터베이스와 Tailwind CSS를 사용하는 풀스택 웹 앱입니다."}

VibeIndex.ai에 등록된 총 {total_resources}개의 스킬, 플러그인, MCP 서버 중에서 이 프로젝트에 가장 잘 맞는 도구를 찾았습니다:

────────────────────────────────────────

1. {name}

{resource_type}

· ⭐ {stars} {One plain sentence about what this does FOR THE USER's project. NO technical jargon. e.g., "이 프로젝트에서 사용 중인 Supabase 데이터베이스를 더 빠르고 안전하게 만들어줍니다."}

{install_command}

────────────────────────────────────────

2. {name}

{resource_type}

· ⭐ {stars} ...

────────────────────────────────────────

VibeIndex.ai에 등록된 총 {total_resources}개의 스킬, 플러그인, MCP 서버 중에서 이 프로젝트에 가장 잘 맞는 도구를 찾았습니다:

────────────────────────────────────────

1. {name}

{resource_type}

{install_command}

────────────────────────────────────────

2. {name}

{resource_type}

· ⭐ {stars} ...

────────────────────────────────────────

설치

필요한 것만 복사해서 실행하세요:

{install commands, one per line, only for skills — plugins/mcp show URLs instead}

💡 더 많은 도구 탐색 → https://vibeindex.ai

undefined

필요한 것만 복사해서 실행하세요:

{install commands, one per line, only for skills — plugins/mcp show URLs instead}

💡 더 많은 도구 탐색 → https://vibeindex.ai

undefined

Writing style guidelines:

写作风格指南:

SIMPLE LANGUAGE ONLY: Write like you're explaining to a friend, not a developer docs page. No jargon like "RLS 정책", "인덱싱", "쿼리 최적화", "번들 최적화". Instead say what it DOES: "더 빠르게", "더 안전하게", "코드를 깔끔하게", "버그를 줄여줍니다"
Focus on benefit: Don't explain HOW it matches (no "package.json에 X가 있고..."). Just say what the user GETS.
One sentence per recommendation: Each description should be exactly one plain sentence. No bullet points, no technical details.
No match percentages: Do not show match scores or percentages to the user.
No description from API: Do NOT use the raw
```
description
```
or
```
description_ko
```
from the API response. Write your own simple sentence based on what the resource does for THIS project.
Language: Translate everything into the user's detected language. The template above is Korean as a base.

仅使用简单语言: 像给朋友解释一样写作，不要像开发者文档页面。不要使用“RLS 정책”、“인덱싱”、“쿼리 최적화”、“번들 최적화”这类行话。而是说明它的作用：“更快”、“更安全”、“让代码更整洁”、“减少错误”
聚焦收益: 不要解释它如何匹配（不要说“package.json에 X가 있고...”），只说用户能得到什么。
每个推荐一句话: 每个描述必须是一句简单的话，不要用项目符号，不要有技术细节。
不要显示匹配百分比: 不要向用户展示匹配分数或百分比。
不要使用API中的描述: 不要使用API响应中的原始
```
description
```
或
```
description_ko
```
字段。根据该资源对当前项目的作用，自己编写简单的句子。
语言: 将所有内容翻译成检测到的用户语言。上面的模板是韩语示例。

Install commands by type:

按类型区分的安装命令:

skill:

npx skills add {github_owner}/{github_repo} --skill {name}

plugin: See

https://vibeindex.ai/plugins/{github_owner}/{github_repo}/{name}

mcp: See

https://vibeindex.ai/mcp/{github_owner}/{github_repo}

marketplace: See

https://vibeindex.ai/marketplaces/{github_owner}/{github_repo}

skill:

npx skills add {github_owner}/{github_repo} --skill {name}

plugin: 查看

https://vibeindex.ai/plugins/{github_owner}/{github_repo}/{name}

mcp: 查看

https://vibeindex.ai/mcp/{github_owner}/{github_repo}

marketplace: 查看

https://vibeindex.ai/marketplaces/{github_owner}/{github_repo}

API Response Format

API响应格式

The

/api/resources

endpoint returns:

json

{
  "data": [
    {
      "name": "resource-name",
      "resource_type": "skill|mcp|plugin|marketplace",
      "description": "...",
      "stars": 12345,
      "github_owner": "owner",
      "github_repo": "repo",
      "tags": ["tag1", "tag2"]
    }
  ],
  "total": 100
}

The

/api/rising-stars

endpoint returns:

json

{
  "rising": [
    {
      "name": "resource-name",
      "resource_type": "mcp",
      "stars": 5000,
      "stars_today": 120
    }
  ],
  "period": "week"
}

Built by Vibe Index - The Claude Code Ecosystem Directory

/api/resources

端点返回:

json

{
  "data": [
    {
      "name": "resource-name",
      "resource_type": "skill|mcp|plugin|marketplace",
      "description": "...",
      "stars": 12345,
      "github_owner": "owner",
      "github_repo": "repo",
      "tags": ["tag1", "tag2"]
    }
  ],
  "total": 100
}

/api/rising-stars

端点返回:

json

{
  "rising": [
    {
      "name": "resource-name",
      "resource_type": "mcp",
      "stars": 5000,
      "stars_today": 120
    }
  ],
  "period": "week"
}

由 Vibe Index 构建 - Claude Code生态系统目录