E2B

@computesdk/e2b

E2B provider for ComputeSDK - Execute code in secure, isolated E2B sandboxes with full filesystem and terminal support.

Installation

npm install @computesdk/e2b

Setup

1. Get your E2B API key from e2b.dev
2. Set the environment variable:

export E2B_API_KEY=e2b_your_api_key_here

Usage

With ComputeSDK

import { compute } from 'computesdk';
import { e2b } from '@computesdk/e2b';

// Set as default provider
compute.setConfig({ 
  provider: e2b({ apiKey: process.env.E2B_API_KEY }) 
});

// Create sandbox
const sandbox = await compute.sandbox.create({});

// Execute Python code
const result = await sandbox.runCode(`
import pandas as pd
import numpy as np

data = {'A': [1, 2, 3], 'B': [4, 5, 6]}
df = pd.DataFrame(data)
print(df)
print(f"Sum: {df.sum().sum()}")
`);

console.log(result.stdout);
// Output:
//    A  B
// 0  1  4
// 1  2  5
// 2  3  6
// Sum: 21

// Clean up
await compute.sandbox.destroy(sandbox.sandboxId);

Direct Usage

import { e2b } from '@computesdk/e2b';

// Create provider
const provider = e2b({ 
  apiKey: 'e2b_your_api_key',
  timeout: 600000 // 10 minutes
});

// Use with compute singleton
const sandbox = await compute.sandbox.create({ provider });

Configuration

Environment Variables

export E2B_API_KEY=e2b_your_api_key_here

Configuration Options

interface E2BConfig {
  /** E2B API key - if not provided, will use E2B_API_KEY env var */
  apiKey?: string;
  /** Default runtime environment */
  runtime?: 'python' | 'node';
  /** Execution timeout in milliseconds */
  timeout?: number;
}

Features

• ✅ Code Execution - Python and Node.js runtime support
• ✅ Command Execution - Run shell commands in sandbox
• ✅ Filesystem Operations - Full file system access via E2B API
• ✅ Terminal Support - Interactive PTY terminals
• ✅ Auto Runtime Detection - Automatically detects Python vs Node.js
• ✅ Data Science Ready - Pre-installed pandas, numpy, matplotlib, etc.

API Reference

Code Execution

// Execute Python code
const result = await sandbox.runCode(`
import json
data = {"message": "Hello from Python"}
print(json.dumps(data))
`, 'python');

// Execute Node.js code  
const result = await sandbox.runCode(`
const data = { message: "Hello from Node.js" };
console.log(JSON.stringify(data));
`, 'node');

// Auto-detection (based on code patterns)
const result = await sandbox.runCode('print("Auto-detected as Python")');

Command Execution

// List files
const result = await sandbox.runCommand('ls', ['-la']);

// Install packages
const result = await sandbox.runCommand('pip', ['install', 'requests']);

// Run scripts
const result = await sandbox.runCommand('python', ['script.py']);

Filesystem Operations

// Write file
await sandbox.filesystem.writeFile('/tmp/hello.py', 'print("Hello World")');

// Read file
const content = await sandbox.filesystem.readFile('/tmp/hello.py');

// Create directory
await sandbox.filesystem.mkdir('/tmp/data');

// List directory contents
const files = await sandbox.filesystem.readdir('/tmp');

// Check if file exists
const exists = await sandbox.filesystem.exists('/tmp/hello.py');

// Remove file or directory
await sandbox.filesystem.remove('/tmp/hello.py');

Terminal Operations

// Create terminal
const terminal = await sandbox.terminal.create({
  command: 'bash',
  cols: 80,
  rows: 24,
  onData: (data: Uint8Array) => {
    const output = new TextDecoder().decode(data);
    console.log('Terminal output:', output);
  }
});

// Write to terminal
await terminal.write('echo "Hello Terminal!"\n');

// Resize terminal
await terminal.resize(120, 30);

// Kill terminal
await terminal.kill();

// List all terminals
const terminals = await sandbox.terminal.list();

// Get terminal by ID
const existingTerminal = await sandbox.terminal.getById('terminal-id');

Sandbox Management

// Get sandbox info
const info = await sandbox.getInfo();
console.log(info.id, info.provider, info.status);

// Get existing sandbox (reconnect)
const existing = await compute.sandbox.getById(provider, 'sandbox-id');

// Destroy sandbox
await compute.sandbox.destroy(provider, 'sandbox-id');

Runtime Detection

The provider automatically detects the runtime based on code patterns:

Python indicators:

• print( statements
• import statements
• def function definitions
• Python-specific syntax (f", __, etc.)

Default: Node.js for all other cases

Error Handling

try {
  const result = await sandbox.runCode('invalid code');
} catch (error) {
  if (error.message.includes('Missing E2B API key')) {
    console.error('Set E2B_API_KEY environment variable');
  } else if (error.message.includes('Invalid E2B API key format')) {
    console.error('E2B API keys should start with "e2b_"');
  } else if (error.message.includes('authentication failed')) {
    console.error('Check your E2B API key');
  } else if (error.message.includes('quota exceeded')) {
    console.error('E2B usage limits reached');
  } else if (error.message.includes('Syntax error')) {
    console.error('Code has syntax errors');
  }
}

Web Framework Integration

Use with web frameworks via the request handler:

import { handleComputeRequest } from 'computesdk';
import { e2b } from '@computesdk/e2b';

export async function POST(request: Request) {
  return handleComputeRequest({
    request,
    provider: e2b({ apiKey: process.env.E2B_API_KEY })
  });
}

Examples

Data Science Workflow

const sandbox = await compute.sandbox.create({});

// Create project structure
await sandbox.filesystem.mkdir('/analysis');
await sandbox.filesystem.mkdir('/analysis/data');
await sandbox.filesystem.mkdir('/analysis/output');

// Write input data
const csvData = `name,age,city
Alice,25,New York
Bob,30,San Francisco
Charlie,35,Chicago`;

await sandbox.filesystem.writeFile('/analysis/data/people.csv', csvData);

// Process data with Python
const result = await sandbox.runCode(`
import pandas as pd
import matplotlib.pyplot as plt

# Read data
df = pd.read_csv('/analysis/data/people.csv')
print("Data loaded:")
print(df)

# Calculate statistics
avg_age = df['age'].mean()
print(f"\\nAverage age: {avg_age}")

# Create visualization
plt.figure(figsize=(8, 6))
plt.bar(df['name'], df['age'])
plt.title('Age by Person')
plt.xlabel('Name')
plt.ylabel('Age')
plt.savefig('/analysis/output/age_chart.png')
print("\\nChart saved to /analysis/output/age_chart.png")

# Save results
results = {
    'total_people': len(df),
    'average_age': avg_age,
    'cities': df['city'].unique().tolist()
}

import json
with open('/analysis/output/results.json', 'w') as f:
    json.dump(results, f, indent=2)

print("Results saved!")
`);

console.log(result.stdout);

// Read the results
const results = await sandbox.filesystem.readFile('/analysis/output/results.json');
console.log('Analysis results:', JSON.parse(results));

// Check if chart was created
const chartExists = await sandbox.filesystem.exists('/analysis/output/age_chart.png');
console.log('Chart created:', chartExists);

Interactive Terminal Session

const sandbox = await compute.sandbox.create({});

// Create interactive Python terminal
const terminal = await sandbox.terminal.create({
  command: 'python3',
  cols: 80,
  rows: 24,
  onData: (data: Uint8Array) => {
    const output = new TextDecoder().decode(data);
    process.stdout.write(output); // Forward to console
  }
});

// Send Python commands
await terminal.write('import numpy as np\n');
await terminal.write('import pandas as pd\n');
await terminal.write('print("Libraries loaded!")\n');
await terminal.write('data = np.array([1, 2, 3, 4, 5])\n');
await terminal.write('print(f"Mean: {data.mean()}")\n');
await terminal.write('exit()\n');

// Wait for commands to execute
await new Promise(resolve => setTimeout(resolve, 2000));

await terminal.kill();

Machine Learning Pipeline

const sandbox = await compute.sandbox.create({
  options: { timeout: 600000 } // 10 minutes for ML tasks
});

// Create ML project structure
await sandbox.filesystem.mkdir('/ml-project');
await sandbox.filesystem.mkdir('/ml-project/data');
await sandbox.filesystem.mkdir('/ml-project/models');

// Generate and process data
const result = await sandbox.runCode(`
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, r2_score
import joblib

# Generate sample dataset
np.random.seed(42)
X = np.random.randn(1000, 5)
y = X.sum(axis=1) + np.random.randn(1000) * 0.1

# Create DataFrame
feature_names = [f'feature_{i}' for i in range(5)]
df = pd.DataFrame(X, columns=feature_names)
df['target'] = y

print(f"Dataset shape: {df.shape}")
print("\\nDataset info:")
print(df.describe())

# Save dataset
df.to_csv('/ml-project/data/dataset.csv', index=False)
print("\\nDataset saved to /ml-project/data/dataset.csv")

# Split data
X_train, X_test, y_train, y_test = train_test_split(
    df[feature_names], df['target'], test_size=0.2, random_state=42
)

# Train model
model = LinearRegression()
model.fit(X_train, y_train)

# Make predictions
y_pred = model.predict(X_test)

# Evaluate
mse = mean_squared_error(y_test, y_pred)
r2 = r2_score(y_test, y_pred)

print(f"\\nModel Performance:")
print(f"MSE: {mse:.4f}")
print(f"R²: {r2:.4f}")

# Save model
joblib.dump(model, '/ml-project/models/linear_model.pkl')
print("\\nModel saved to /ml-project/models/linear_model.pkl")

# Save results
results = {
    'mse': mse,
    'r2': r2,
    'feature_importance': dict(zip(feature_names, model.coef_)),
    'intercept': model.intercept_
}

import json
with open('/ml-project/results.json', 'w') as f:
    json.dump(results, f, indent=2)

print("Results saved!")
`);

console.log(result.stdout);

// Read the results
const results = await sandbox.filesystem.readFile('/ml-project/results.json');
console.log('ML Results:', JSON.parse(results));

// Verify model file exists
const modelExists = await sandbox.filesystem.exists('/ml-project/models/linear_model.pkl');
console.log('Model saved:', modelExists);

Best Practices

1. Resource Management: Always destroy sandboxes when done to free resources
2. Error Handling: Use try-catch blocks for robust error handling
3. Timeouts: Set appropriate timeouts for long-running tasks
4. File Organization: Use the filesystem API to organize project files
5. Terminal Sessions: Clean up terminal sessions with terminal.kill()
6. API Key Security: Never commit API keys to version control

Limitations

• Memory Limits: Subject to E2B sandbox memory constraints
• Network Access: Limited outbound network access
• File Persistence: Files are not persisted between sandbox sessions
• Execution Time: Subject to E2B timeout limits

Support

• E2B Documentation
• ComputeSDK Issues
• E2B Support