Pistachio MCP Server
A remote MCP server built with Node.js and TypeScript that enables tool calls and prompt templates via streamable HTTP transport. It includes example implementations for a calculator and localized greetings, featuring built-in CORS support for web-based clients.
README
Pistachio MCP Server
A remote MCP (Model Context Protocol) server built with Node.js and TypeScript for mobile app development. This server provides tools and prompts for remote project management, testing, and asset search.
Prerequisites
Before setting up this repository, ensure you have the following installed:
- Node.js (v20 or higher)
- Yarn (v4.10.3 or compatible version)
- Git
Setup Instructions
1. Clone the Repository
git clone <repository-url>
cd pistachio-mcp
2. Install Yarn (if not already installed)
This project uses Yarn 4.10.3 as the package manager. If you don't have Yarn installed or need to upgrade:
# Install Yarn globally (if needed)
npm install -g yarn
# Or use Corepack (recommended for Node.js 16.10+)
corepack enable
corepack prepare yarn@4.10.3 --activate
3. Install Dependencies
Install all project dependencies using Yarn:
yarn install
This will install all dependencies listed in package.json, including:
@modelcontextprotocol/sdk- MCP SDKfirebase- Firebase SDK for Firestore and Auth@google-cloud/storage- Google Cloud Storage clientzod- Schema validation- TypeScript and development dependencies
4. Configure Environment Variables
Create a .env file in the root directory (if needed for custom configuration):
# Optional: Set custom port (default: 3001)
PORT=3001
# Optional: Set number of worker threads (default: 2)
NUM_WORKERS=2
# Optional: Set Node environment
NODE_ENV=development
# Optional: Firebase emulator configuration (for development)
FIREBASE_AUTH_EMULATOR_HOST=localhost:9099
FIRESTORE_EMULATOR_HOST=localhost:8080
# Optional: Google Cloud Storage bucket for weekly expiring assets
GCS_BUCKET_WEEKLY_EXPIRING=dev-pistachio-assets-weekly-expiring
Note: The project uses hardcoded Firebase configuration in src/utils/ServerStorageUtils.ts. For production, you may want to move this to environment variables.
5. Configure Google Cloud Storage (Optional)
If you plan to use Google Cloud Storage features (image uploads), you'll need to set up authentication:
- Create a service account in Google Cloud Console
- Download the service account key as a JSON file
- Set the environment variable:
export GOOGLE_APPLICATION_CREDENTIALS="/path/to/your/service-account-key.json"
Alternatively, if running on Google Cloud Platform, authentication is handled automatically.
6. Build the Project
Compile TypeScript to JavaScript:
yarn build
This will generate the dist/ directory with compiled JavaScript files.
7. Verify Installation
Run the tests to verify everything is set up correctly:
yarn test:run
Running the Server
Development Mode
Run the server in development mode with hot reload:
yarn dev
The server will start on port 3001 by default (or the port specified by the PORT environment variable).
Production Mode
Build and run the compiled server:
yarn build
node dist/index.js
Direct Execution (Development)
Run TypeScript directly without building:
yarn start
Server Configuration
Default Settings
- Port: 3001 (configurable via
PORTenvironment variable) - Workers: 2 (configurable via
NUM_WORKERSenvironment variable) - Host:
0.0.0.0(accessible from remote clients)
MCP Endpoint
The server exposes an MCP endpoint at /message that accepts:
- GET: Establish SSE stream for receiving messages
- POST: Send MCP requests
- OPTIONS: CORS preflight requests
Example endpoint: http://localhost:3001/message
Available Tools
The server provides the following MCP tools:
search_image- Search for imagessearch_icon- Search for iconscreate_remote_project- Create a remote projectremote_kdoctor- Run kdoctor diagnosticsremote_clean_project- Clean a remote projectremote_test_android- Run Android tests remotely
Available Prompts
The server provides the following MCP prompts:
create_pistachio_project- Generate instructions for creating a Pistachio projectstart_sync- Generate instructions for starting sync
Development
Running Tests
# Run tests in watch mode
yarn test
# Run tests once
yarn test:run
Linting
# Check for linting errors
yarn lint
# Fix linting errors automatically
yarn lint:fix
TypeScript Compilation Errors
If you encounter TypeScript errors:
# Clean and rebuild
rm -rf dist
yarn build
License
MIT
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。