context7-auto-research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Context7 Auto Research Skill

Context7 Auto Research 技能

This skill automatically fetches current documentation from Context7 API when detecting library/framework-related queries, ensuring responses use up-to-date information instead of potentially outdated training data.

当检测到与库/框架相关的查询时，该Skill会自动从Context7 API获取最新文档，确保回复使用当前信息而非可能过时的训练数据。

Automatic Activation Triggers

自动激活触发条件

This skill should activate proactively when the user's message contains:

当用户消息包含以下内容时，该Skill应主动激活：

Implementation Queries (实现相关)

实现相关查询

"如何实现" / "怎么写" / "怎么做"
"How do I..." / "How to..." / "How can I..."
"Show me how to..." / "Write code for..."

"如何实现" / "怎么写" / "怎么做"
"How do I..." / "How to..." / "How can I..."
"Show me how to..." / "Write code for..."

Configuration & Setup (配置相关)

配置与设置相关

"配置" / "设置" / "安装"
"configure" / "setup" / "install"
"初始化" / "initialize"

"配置" / "设置" / "安装"
"configure" / "setup" / "install"
"初始化" / "initialize"

Documentation Requests (文档相关)

文档相关请求

"文档" / "参考" / "API"
"documentation" / "docs" / "reference"
"查看" / "look up"

"文档" / "参考" / "API"
"documentation" / "docs" / "reference"
"查看" / "look up"

Library/Framework Mentions (库/框架提及)

库/框架提及

React, Vue, Angular, Svelte, Solid
Next.js, Nuxt, Remix, Astro
Express, Fastify, Koa, Hono
Prisma, Drizzle, TypeORM
Supabase, Firebase, Clerk
Tailwind, shadcn/ui, Radix
Any npm package or GitHub repository

React, Vue, Angular, Svelte, Solid
Next.js, Nuxt, Remix, Astro
Express, Fastify, Koa, Hono
Prisma, Drizzle, TypeORM
Supabase, Firebase, Clerk
Tailwind, shadcn/ui, Radix
任何npm包或GitHub仓库

Code Generation Requests (代码生成)

代码生成请求

"生成代码" / "写一个" / "创建"
"generate" / "create" / "build"
"implement" / "add feature"

"生成代码" / "写一个" / "创建"
"generate" / "create" / "build"
"implement" / "add feature"

Research Process

研究流程

When triggered, follow this workflow:

触发后，遵循以下工作流：

Step 1: Extract Library Information

步骤1：提取库信息

Identify the library/framework from the user's query:

Library name (e.g., "react", "next.js", "prisma")
Version if specified (e.g., "React 19", "Next.js 15")
Specific feature/API mentioned (e.g., "useEffect", "middleware", "relations")

从用户查询中识别出库/框架：

库名称（例如："react", "next.js", "prisma"）
若指定版本（例如："React 19", "Next.js 15"）
提及的特定功能/API（例如："useEffect", "middleware", "relations"）

Step 2: Search for Library

步骤2：搜索库

Use Task tool to call context7-fetcher sub-skill:

Task parameters:
- subagent_type: Bash
- description: "Search Context7 for library"
- prompt: node .claude/skills/context7-auto-research/context7-api.js search "<library-name>" "<user-query>"

Example:

Task: Search for Next.js
Prompt: node .claude/skills/context7-auto-research/context7-api.js search "next.js" "How to configure middleware in Next.js 15"

Response format:

json

{
  "libraries": [
    {
      "id": "/vercel/next.js",
      "name": "Next.js",
      "description": "The React Framework",
      "trustScore": 95,
      "versions": ["v15.1.8", "v14.2.0", "v13.5.0"]
    }
  ]
}

Why use Task tool?

Uses
```
context: fork
```
from context7-fetcher sub-skill
Avoids carrying conversation history to API calls
Reduces Token consumption

使用Task工具调用context7-fetcher子Skill：

Task parameters:
- subagent_type: Bash
- description: "Search Context7 for library"
- prompt: node .claude/skills/context7-auto-research/context7-api.js search "<library-name>" "<user-query>"

示例：

Task: Search for Next.js
Prompt: node .claude/skills/context7-auto-research/context7-api.js search "next.js" "How to configure middleware in Next.js 15"

响应格式：

json

{
  "libraries": [
    {
      "id": "/vercel/next.js",
      "name": "Next.js",
      "description": "The React Framework",
      "trustScore": 95,
      "versions": ["v15.1.8", "v14.2.0", "v13.5.0"]
    }
  ]
}

为何使用Task工具？

使用context7-fetcher子Skill的
```
context: fork
```
避免将对话历史带入API调用
减少Token消耗

Step 3: Select Best Match

步骤3：选择最佳匹配

From search results, choose the library based on:

Exact name match to user's query
Highest trust score (indicates quality/popularity)
Version match if user specified (e.g., "Next.js 15" → prefer v15.x)
Official packages over community forks

从搜索结果中，根据以下条件选择库：

名称精确匹配用户查询
信任分数最高（代表质量/流行度）
若用户指定版本则版本匹配（例如："Next.js 15" → 优先选择v15.x）
官方包优先于社区分支

Step 4: Fetch Documentation

步骤4：获取文档

Use Task tool to call context7-fetcher sub-skill:

Task parameters:
- subagent_type: Bash
- description: "Fetch documentation from Context7"
- prompt: node .claude/skills/context7-auto-research/context7-api.js context "<library-id>" "<specific-query>"

Example:

Task: Fetch Next.js middleware docs
Prompt: node .claude/skills/context7-auto-research/context7-api.js context "/vercel/next.js" "middleware configuration"

Response format:

json

{
  "results": [
    {
      "title": "Middleware",
      "content": "Middleware allows you to run code before a request is completed...",
      "source": "docs/app/building-your-application/routing/middleware.md",
      "relevance": 0.95
    }
  ]
}

Why use Task tool?

Independent context for API calls
No conversation history overhead
Faster execution

使用Task工具调用context7-fetcher子Skill：

Task parameters:
- subagent_type: Bash
- description: "Fetch documentation from Context7"
- prompt: node .claude/skills/context7-auto-research/context7-api.js context "<library-id>" "<specific-query>"

示例：

Task: Fetch Next.js middleware docs
Prompt: node .claude/skills/context7-auto-research/context7-api.js context "/vercel/next.js" "middleware configuration"

响应格式：

json

{
  "results": [
    {
      "title": "Middleware",
      "content": "Middleware allows you to run code before a request is completed...",
      "source": "docs/app/building-your-application/routing/middleware.md",
      "relevance": 0.95
    }
  ]
}

为何使用Task工具？

API调用使用独立上下文
无对话历史开销
执行速度更快

Step 5: Integrate into Response

步骤5：整合到回复中

Use the fetched documentation to:

Answer accurately with current information
Include code examples from the docs
Cite version when relevant
Provide context about the feature/API

使用获取到的文档：

准确回答，使用当前信息
包含代码示例，来自文档
相关时注明版本
提供功能/API的上下文说明

Helper Script Usage

辅助脚本使用

The

context7-api.js

script provides two commands:

context7-api.js

脚本提供两个命令：

Search Library

搜索库

bash

node context7-api.js search <libraryName> <query>

Returns matching libraries with metadata
Use for initial library resolution

bash

node context7-api.js search <libraryName> <query>

返回匹配的库及元数据
用于初始库识别

Get Context

获取上下文

bash

node context7-api.js context <libraryId> <query>

Returns relevant documentation snippets
Use after selecting a library

bash

node context7-api.js context <libraryId> <query>

返回相关文档片段
选择库后使用

Environment Setup

环境设置

The script supports two ways to configure the API key:

脚本支持两种配置API密钥的方式：

Option 1: .env File (Recommended)

选项1：.env文件（推荐）

Create a

.env

file in the skill directory:

bash

undefined

在技能目录中创建

.env

文件：

bash

undefined

In .claude/skills/context7-auto-research/.env

CONTEXT7_API_KEY=your_api_key_here


You can copy from the example:

```bash
cp .env.example .env

CONTEXT7_API_KEY=your_api_key_here


你可以从示例文件复制：

```bash
cp .env.example .env

Then edit .env with your actual API key

然后编辑.env文件填入实际API密钥

undefined

undefined

Option 2: Environment Variable

选项2：环境变量

bash

export CONTEXT7_API_KEY="your-api-key"

Priority: Environment variable > .env file

Get API Key: Visit context7.com/dashboard to register and obtain your API key.

If not set, the API will use public rate limits (lower quota).

bash

export CONTEXT7_API_KEY="your-api-key"

优先级： 环境变量 > .env文件

获取API密钥： 访问context7.com/dashboard注册并获取API密钥。

若未设置，API将使用公共速率限制（配额较低）。

Best Practices

最佳实践

Query Specificity

查询特异性

Pass the full user question as the query parameter for better relevance
Include specific feature names (e.g., "useEffect cleanup" vs just "useEffect")

将完整用户问题作为查询参数，以获得更好的相关性
包含特定功能名称（例如："useEffect cleanup" 而非仅 "useEffect"）

Version Awareness

版本意识

When users mention versions, use version-specific library IDs
Example:
```
/vercel/next.js/v15.1.8
```
instead of
```
/vercel/next.js
```

当用户提及版本时，使用特定版本的库ID
示例：
```
/vercel/next.js/v15.1.8
```
而非
```
/vercel/next.js
```

Error Handling

错误处理

If library search returns no results, inform user and suggest alternatives
If API fails, fall back to training data but mention it may be outdated
Handle rate limits gracefully (429 errors)

若库搜索无结果，告知用户并建议替代方案
若API调用失败，回退到训练数据但注明可能过时
优雅处理速率限制（429错误）

Response Quality

回复质量

Don't dump entire documentation - extract relevant parts
Combine multiple doc snippets if needed for complete answer
Always include practical code examples

不要直接输出整个文档 - 提取相关部分
若需要完整回答，可组合多个文档片段
始终包含实用代码示例

Example Workflows

示例工作流

Example 1: React Hook Question

示例1：React Hook问题

User: "How do I use useEffect to fetch data in React 19?"

Skill Actions:

Detect trigger: "How do I use" + "useEffect" + "React 19"

Search:

node context7-api.js search "react" "useEffect fetch data"

Select:
```
/facebook/react/v19.0.0
```
(version match)

Fetch:

node context7-api.js context "/facebook/react/v19.0.0" "useEffect data fetching"

Respond with current React 19 patterns (e.g., using
```
use()
```
hook if applicable)

用户： "How do I use useEffect to fetch data in React 19?"

Skill操作：

检测触发条件："How do I use" + "useEffect" + "React 19"

搜索：

node context7-api.js search "react" "useEffect fetch data"

选择：
```
/facebook/react/v19.0.0
```
（版本匹配）

获取：

node context7-api.js context "/facebook/react/v19.0.0" "useEffect data fetching"

使用当前React 19模式回复（例如：若适用则使用
```
use()
```
hook）

Example 2: Next.js Configuration

示例2：Next.js配置

User: "配置 Next.js 15 的中间件"

Skill Actions:

Detect trigger: "配置" + "Next.js 15" + "中间件"

Search:

node context7-api.js search "next.js" "middleware configuration"

Select:
```
/vercel/next.js/v15.1.8
```

Fetch:

node context7-api.js context "/vercel/next.js/v15.1.8" "middleware"

Respond with Next.js 15 middleware setup

用户： "配置 Next.js 15 的中间件"

Skill操作：

检测触发条件："配置" + "Next.js 15" + "中间件"

搜索：

node context7-api.js search "next.js" "middleware configuration"

选择：
```
/vercel/next.js/v15.1.8
```

获取：

node context7-api.js context "/vercel/next.js/v15.1.8" "middleware"

回复Next.js 15中间件设置方法

Example 3: Prisma Relations

示例3：Prisma关联

User: "Show me how to define one-to-many relations in Prisma"

Skill Actions:

Detect trigger: "Show me how" + "Prisma"

Search:

node context7-api.js search "prisma" "one-to-many relations"

Select:
```
/prisma/prisma
```
(highest trust score)

Fetch:

node context7-api.js context "/prisma/prisma" "one-to-many relations"

Respond with Prisma schema examples

用户： "Show me how to define one-to-many relations in Prisma"

Skill操作：

检测触发条件："Show me how" + "Prisma"

搜索：

node context7-api.js search "prisma" "one-to-many relations"

选择：
```
/prisma/prisma
```
（信任分数最高）

获取：

node context7-api.js context "/prisma/prisma" "one-to-many relations"

回复Prisma schema示例

Architecture: Context Separation

架构：上下文分离

Why Split into Two Skills?

为何拆分为两个Skill？

This skill adopts a two-phase architecture:

Main Skill (context7-auto-research) - Needs conversation context:
- Detect trigger keywords in user message
- Extract user query intent
- Select best matching library (version, name, trust score)
- Integrate documentation into response
Sub-Skill (context7-fetcher) - Independent context (
```
context: fork
```
):
- Execute API calls to Context7
- Pure HTTP requests, no conversation history needed
- Reduce Token consumption

该Skill采用两阶段架构：

主Skill（context7-auto-research） - 需要对话上下文：
- 检测用户消息中的触发关键词
- 提取用户查询意图
- 选择最佳匹配库（版本、名称、信任分数）
- 将文档整合到回复中
子Skill（context7-fetcher） - 独立上下文（
```
context: fork
```
）：
- 执行Context7 API调用
- 纯HTTP请求，无需对话历史
- 减少Token消耗

Benefits

优势

Aspect	Main Skill	Sub-Skill
Context	Full conversation	Fork (independent)
Purpose	Intent analysis	API execution
Token usage	Higher	Lower
Execution	Sequential	Can be parallel

维度	Main Skill	Sub-Skill
上下文	完整对话上下文	Fork（独立上下文）
用途	意图分析	API执行
Token消耗	较高	较低
执行方式	顺序执行	可并行执行

Call Flow

调用流程

User Query → Main Skill (detect + analyze)
                ↓
           Task Tool → Sub-Skill (API search)
                ↓
           Main Skill (select best match)
                ↓
           Task Tool → Sub-Skill (API fetch docs)
                ↓
           Main Skill (integrate + respond)

用户查询 → 主Skill（检测+分析）
                ↓
           Task工具 → 子Skill（API搜索）
                ↓
           主Skill（选择最佳匹配）
                ↓
           Task工具 → 子Skill（API获取文档）
                ↓
           主Skill（整合+回复）

Integration with Existing Skills

与现有Skill的集成

This skill complements the existing

documentation-lookup

skill:

auto-research: Proactive, automatic activation
documentation-lookup: Manual, user-invoked via
```
/context7:docs
```

Both can coexist - use auto-research for seamless UX, documentation-lookup for explicit queries.

该Skill补充现有

documentation-lookup

Skill：

auto-research：主动、自动激活
documentation-lookup：手动、用户通过
```
/context7:docs
```
调用

两者可共存 - auto-research用于无缝用户体验，documentation-lookup用于明确查询。

Performance Considerations

性能注意事项

Cache responses: Documentation changes infrequently
Parallel requests: If user asks about multiple libraries, fetch in parallel using multiple Task calls
Timeout handling: Set reasonable timeouts (5-10s) for API calls
Fallback strategy: If API unavailable, use training data with disclaimer
Context efficiency: Sub-skill uses fork context to minimize Token consumption

缓存响应：文档变更频率低
并行请求：若用户询问多个库，使用多个Task调用并行获取
超时处理：为API调用设置合理超时（5-10秒）
回退策略：若API不可用，使用训练数据并附带免责声明
上下文效率：子Skill使用fork上下文以最小化Token消耗

Limitations

局限性

Requires internet connection for API access
Subject to Context7 API rate limits
May not have documentation for very new or obscure libraries
Documentation quality depends on source repository structure

需要互联网连接以访问API
受Context7 API速率限制约束
可能没有极新或小众库的文档
文档质量取决于源仓库结构 ",