E2B MCP Server
Enables AI assistants to securely execute Python and JavaScript code in sandboxed environments, with file management and package installation.
README
E2B MCP Server
A production-ready Model Context Protocol (MCP) server that integrates AI assistants with E2B's sandboxed code execution environment.
Overview
This MCP server provides secure, sandboxed code execution capabilities for Python and JavaScript, along with file management and package installation features. Built for the Klavis AI coding assignment, it demonstrates enterprise-grade architecture with comprehensive error handling, security validation, and resource management.
Features
Core Tools
execute_python- Execute Python code in sandboxed environmentexecute_javascript- Execute JavaScript/Node.js codecreate_file- Create files in sandboxread_file- Read files from sandboxlist_files- List directory contentsinstall_packages- Install Python (pip) or Node.js (npm) packagesget_sandbox_info- Get sandbox status and resource information
Key Features
- Security-First Design: Input validation, output sanitization, and dangerous pattern detection
- Resource Management: Automatic sandbox cleanup, idle timeout handling, and resource monitoring
- Production Ready: Comprehensive logging, error handling, and graceful shutdown
- Multi-language Support: Both Python and JavaScript execution environments
- Persistent Sessions: Reuse sandboxes across tool calls for better performance
Installation
-
Clone and Install Dependencies
git clone <repository> cd my-mcp-server npm install -
Set Environment Variables
export E2B_API_KEY="your-e2b-api-key" export LOG_LEVEL="info" # Optional: debug, info, warn, error export NODE_ENV="production" # Optional: enables file logging -
Build the Project
npm run build -
Test the Installation
npm test
Usage
As MCP Server
Configure your MCP client to use this server:
{
"mcpServers": {
"e2b-server": {
"command": "node",
"args": ["/path/to/my-mcp-server/dist/index.js"],
"env": {
"E2B_API_KEY": "your-api-key"
}
}
}
}
Development Mode
npm run dev
Direct Testing
# Run comprehensive tests
npm test
# Or run the built server directly
npm start
API Reference
execute_python
Execute Python code in a sandboxed environment.
Parameters:
code(string, required): Python code to executesandbox_id(string, optional): Specific sandbox ID to reuse
Example:
print("Hello from E2B!")
import numpy as np
data = np.array([1, 2, 3, 4, 5])
print(f"Mean: {np.mean(data)}")
execute_javascript
Execute JavaScript/Node.js code in a sandboxed environment.
Parameters:
code(string, required): JavaScript code to executesandbox_id(string, optional): Specific sandbox ID to reuse
Example:
console.log("Hello from Node.js!");
const fs = require('fs');
const data = [1, 2, 3, 4, 5];
const mean = data.reduce((a, b) => a + b) / data.length;
console.log(`Mean: ${mean}`);
create_file
Create a file in the sandbox environment.
Parameters:
path(string, required): File path to createcontent(string, required): File contentsandbox_id(string, optional): Specific sandbox ID
Example:
{
"path": "data/example.txt",
"content": "Hello, E2B MCP Server!"
}
read_file
Read a file from the sandbox environment.
Parameters:
path(string, required): File path to readsandbox_id(string, optional): Specific sandbox ID
list_files
List files in a directory.
Parameters:
path(string, optional): Directory path (defaults to current directory)sandbox_id(string, optional): Specific sandbox ID
install_packages
Install packages in the sandbox environment.
Parameters:
packages(array of strings, required): Package names to installlanguage(string, required): "python" or "javascript"sandbox_id(string, optional): Specific sandbox ID
Examples:
{
"packages": ["numpy", "pandas", "matplotlib"],
"language": "python"
}
{
"packages": ["lodash", "axios", "moment"],
"language": "javascript"
}
get_sandbox_info
Get information about sandbox status and resource usage.
Parameters:
sandbox_id(string, optional): Specific sandbox ID (if not provided, lists all sandboxes)
Security Features
Input Validation
- Code length limits (50KB max)
- Dangerous pattern detection
- Package name validation
- File path sanitization
Output Sanitization
- ANSI escape code removal
- Secret detection and masking
- Output length limits (10KB max)
Resource Limits
- Execution timeout (30 seconds)
- Idle sandbox cleanup (5 minutes)
- Maximum file size (10MB)
- Package installation limits
Dangerous Pattern Detection
The server monitors for potentially dangerous patterns:
- Network operations (
socket,requests,fetch) - System file access (
/etc,/root,/sys) - Process operations (
subprocess,child_process) - Secret patterns (API keys, passwords)
Architecture
Core Components
-
SandboxService (
src/sandbox-manager.ts)- Manages E2B sandbox lifecycle
- Handles resource cleanup and idle timeouts
- Provides sandbox pooling and reuse
-
E2BTools (
src/tools.ts)- Implements all MCP tool handlers
- Integrates with E2B API
- Handles execution and file operations
-
SecurityValidator (
src/security.ts)- Input validation and sanitization
- Pattern detection for dangerous code
- Output sanitization and secret masking
-
Logger (
src/logger.ts)- Structured logging with Winston
- Development and production configurations
- Security event logging
Error Handling
- Comprehensive try-catch blocks
- Graceful degradation
- Detailed error logging
- User-friendly error messages
- Automatic resource cleanup on failures
Resource Management
- Automatic sandbox cleanup on idle timeout
- Graceful shutdown handlers
- Memory and resource monitoring
- Connection pooling for better performance
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
E2B_API_KEY |
Yes | - | Your E2B API key |
LOG_LEVEL |
No | info | Logging level (debug, info, warn, error) |
NODE_ENV |
No | development | Environment (enables file logging in production) |
Development
Project Structure
src/
├── index.ts # Main MCP server entry point
├── sandbox-manager.ts # E2B sandbox lifecycle management
├── tools.ts # MCP tool implementations
├── security.ts # Security validation and sanitization
├── logger.ts # Logging configuration
├── types.ts # TypeScript type definitions
└── test.ts # Comprehensive test suite
Building
npm run build
Testing
# Mock tests (no API key required) - verify implementation structure
npm run test-mock
# Full functionality tests (requires E2B_API_KEY environment variable)
npm test
Mock Testing:
The test-mock script verifies that your implementation is structurally correct without requiring an E2B API key. This is useful for:
- Verifying tool definitions are correct
- Checking method signatures
- Testing implementation structure
- CI/CD environments where API keys aren't available
Full Testing:
The test script runs comprehensive functionality tests including:
- Python code execution
- JavaScript code execution
- File operations (create, read, list)
- Package installation (Python & JavaScript)
- Sandbox management
- Error handling
Linting and Type Checking
npx tsc --noEmit
Production Deployment
Docker Deployment
Create a Dockerfile:
FROM node:18-alpine
WORKDIR /app
COPY package*.json ./
RUN npm ci --only=production
COPY dist/ ./dist/
ENV NODE_ENV=production
CMD ["npm", "start"]
Environment Setup
# Production environment variables
export NODE_ENV=production
export LOG_LEVEL=info
export E2B_API_KEY=your-production-api-key
Monitoring
The server provides structured JSON logging suitable for log aggregation services:
{
"level": "info",
"message": "Executing Python code in sandbox abc-123",
"timestamp": "2024-01-01T12:00:00.000Z",
"service": "e2b-mcp-server"
}
Performance Considerations
- Sandbox Reuse: Sandboxes are reused across tool calls to reduce latency
- Idle Cleanup: Automatic cleanup after 5 minutes of inactivity
- Output Limits: Output is truncated at 10KB to prevent memory issues
- Timeout Management: 30-second execution timeout with graceful error handling
- Resource Monitoring: Built-in sandbox resource usage tracking
Troubleshooting
Common Issues
-
"E2B_API_KEY environment variable is required"
- Ensure your E2B API key is set in environment variables
- Verify the key is valid and has appropriate permissions
-
"Sandbox creation failed"
- Check E2B service status
- Verify API key permissions
- Check network connectivity
-
"Security validation failed"
- Review code for dangerous patterns
- Check file paths for directory traversal
- Verify package names follow naming conventions
-
"Execution timeout"
- Code took longer than 30 seconds to execute
- Consider optimizing code or breaking into smaller chunks
Fixed Issues (Latest Update)
✅ Code Execution Issue Resolved
The previous issue with Jupyter kernel port (49999) not being open has been completely resolved. The server now uses E2B's latest v1.5.1 SDK which provides direct code execution without relying on Jupyter kernels.
What was fixed:
- Updated from legacy Jupyter-based execution to E2B's modern code execution API
- Fixed
execute_python✅ - Now works with E2B v1.5.1 direct execution - Fixed
execute_javascript✅ - Now works with proper Node.js wrapping - Fixed
install_packages✅ - Now works for both Python (pip) and JavaScript (npm)
All features now working:
- ✅
execute_python- Execute Python code in sandboxed environment - ✅
execute_javascript- Execute JavaScript/Node.js code - ✅
create_file- Create files in sandbox - ✅
read_file- Read files from sandbox - ✅
list_files- List directory contents - ✅
install_packages- Install Python (pip) or Node.js (npm) packages - ✅
get_sandbox_info- Get sandbox status and resource information
Logging
Enable debug logging for detailed troubleshooting:
export LOG_LEVEL=debug
npm start
Contributing
This project follows TypeScript strict mode and includes comprehensive error handling. When contributing:
- Follow the existing code style
- Add appropriate error handling
- Include security validation for new features
- Update tests for new functionality
- Add logging for important operations
License
MIT License - see LICENSE file for details.
Support
For issues specific to this MCP server implementation, please check the troubleshooting section above. For E2B-related issues, consult the E2B documentation.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.