Grep App MCP Server
Enables powerful code search across millions of public GitHub repositories using grep.app API, with capabilities to discover code examples, retrieve specific files, and batch fetch multiple files for learning from open source projects.
README
Grep App MCP Server
A Model Context Protocol (MCP) server that provides powerful code search capabilities across public GitHub repositories using the grep.app API. Perfect for code discovery, learning from open source projects, and finding implementation examples.
🚀 Features
- 🔍 Advanced Code Search: Search across millions of public repositories on GitHub
- 📁 File Retrieval: Fetch specific files or batches of files from GitHub
- 🎯 Flexible Filtering: Filter by language, repository, file path, and more
- 📊 Multiple Output Formats: JSON, numbered lists, or formatted text
- ⚡ Batch Operations: Retrieve multiple files efficiently in parallel
- 🔄 Result Caching: Cache search results for quick file retrieval
- 📝 Comprehensive Logging: Built-in logging with daily rotation
🛠️ Installation & Setup
Prerequisites
- Node.js 18+
- npm or yarn
- Claude Code CLI
Quick Start
-
Clone or download this repository
git clone https://github.com/ai-tools-all/grep_app_mcp.git cd grep_app_mcp -
Install dependencies
npm install -
Build the project
npm run build -
Test the server
# HTTP mode (recommended for development) ./run.sh http dev # or STDIO mode ./run.sh stdio dev
🔧 Adding to Claude Code
Method 1: Using MCP Configuration
Add this server to your Claude Code MCP configuration:
{
"mcpServers": {
"grep_app": {
"command": "node",
"args": ["/path/to/grep_app_mcp/dist/server-stdio.js"],
"env": {}
}
}
}
Method 2: Using HTTP Transport
For HTTP mode, add to your configuration:
{
"mcpServers": {
"grep_app": {
"url": "http://localhost:8603/mcp"
}
}
}
Then start the server:
./run.sh http prod
📖 run.sh Usage
The run.sh script provides convenient ways to start the server:
Basic Usage
./run.sh [mode] [environment]
Modes
http- HTTP server with streaming support (default)stdio- STDIO server for direct MCP integration
Environments
dev- Development mode with hot reload (default)prod- Production mode (requires build step)
Examples
# Development (default) - HTTP server with hot reload
./run.sh
# Development - HTTP server
./run.sh http dev
# Production - HTTP server
./run.sh http prod
# Development - STDIO server
./run.sh stdio dev
# Production - STDIO server
./run.sh stdio prod
# Show help
./run.sh help
HTTP Endpoints (when using HTTP mode)
- HTTP Streaming:
http://localhost:8603/mcp - Server-Sent Events:
http://localhost:8603/sse
🔨 Available Tools
1. searchCode
Search for code across public GitHub repositories.
Parameters:
query(required) - Search query stringjsonOutput- Return JSON format (default: false)numberedOutput- Return numbered list format (default: false)caseSensitive- Case-sensitive searchuseRegex- Treat query as regex patternwholeWords- Search whole words onlyrepoFilter- Filter by repository name patternpathFilter- Filter by file path patternlangFilter- Filter by programming language(s)
Example:
{
"query": "async function fetchData",
"langFilter": "TypeScript,JavaScript",
"numberedOutput": true
}
2. github_file
Fetch a single file from a GitHub repository.
Parameters:
owner(required) - Repository ownerrepo(required) - Repository namepath(required) - File pathref(optional) - Branch/commit/tag reference
Example:
{
"owner": "microsoft",
"repo": "vscode",
"path": "src/vs/editor/editor.api.ts"
}
3. github_batch_files
Fetch multiple files from GitHub repositories in parallel.
Parameters:
files(required) - Array of file objects with owner, repo, path, and optional ref
Example:
{
"files": [
{"owner": "facebook", "repo": "react", "path": "packages/react/index.js"},
{"owner": "microsoft", "repo": "TypeScript", "path": "src/compiler/types.ts"}
]
}
4. batch_retrieve_files
Retrieve files from previously cached search results.
Parameters:
query(required) - Original search queryresultNumbers(optional) - Array of result indices to retrieve
Example:
{
"query": "tower_governor",
"resultNumbers": [1, 2, 3]
}
🎯 Common Workflows
1. Code Discovery
# Search for React hooks examples
searchCode("useEffect cleanup", langFilter: "JavaScript,TypeScript")
# Retrieve specific files from results
batch_retrieve_files(query: "useEffect cleanup", resultNumbers: [1, 3, 5])
2. Learning Patterns
# Find authentication implementations
searchCode("JWT authentication middleware", repoFilter: "*express*")
# Get specific implementation details
github_file(owner: "auth0", repo: "express-jwt", path: "lib/index.js")
3. API Research
# Discover API patterns
searchCode("GraphQL resolver", pathFilter: "*/resolvers/*")
# Compare multiple implementations
github_batch_files([
{owner: "apollographql", repo: "apollo-server", path: "packages/apollo-server-core/src/resolvers.ts"},
{owner: "graphql", repo: "graphql-js", path: "src/execution/execute.js"}
])
📋 Development
Available Scripts
npm run build- Build TypeScript to JavaScriptnpm run start- Start production HTTP servernpm run start-stdio- Start production STDIO servernpm run dev- Start development HTTP server with hot reloadnpm run dev-stdio- Start development STDIO server with hot reloadnpm run test-client- Run test client
Project Structure
src/
├── core/ # Core utilities (logger, types)
├── tools/ # MCP tool implementations
├── server.ts # HTTP server entry point
└── server-stdio.ts # STDIO server entry point
📝 Logging
The server includes comprehensive logging with daily rotation:
- Location:
logs/directory - Rotation: Daily with date-based filenames
- Levels: error, warn, info, debug
- Format: JSON with timestamps
🤝 Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Add tests if applicable
- Submit a pull request
📄 License
ISC License - see package.json for details
🔗 Related
- grep.app - The search service powering this tool
- Model Context Protocol - MCP specification
- Claude Code - Claude's official CLI
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。