@computesdk/vercel

Vercel provider for ComputeSDK - Execute Node.js and Python code in secure, isolated Vercel sandboxes.

Installation

npm install @computesdk/vercel

Authentication

Vercel provider supports two authentication methods:

Method 1: OIDC Token (Recommended)

The simplest way to authenticate. Vercel manages token expiration automatically.

Development:

vercel env pull  # Downloads VERCEL_OIDC_TOKEN to .env.local

Production: Vercel automatically provides VERCEL_OIDC_TOKEN in your deployment environment.

Method 2: Access Token + Team/Project IDs

Alternative method using explicit credentials:

export VERCEL_TOKEN=your_vercel_token_here
export VERCEL_TEAM_ID=your_team_id_here
export VERCEL_PROJECT_ID=your_project_id_here

Get your token from Vercel Account Tokens

Quick Start

Gateway Mode (Recommended)

Use the gateway for zero-config auto-detection:

import { compute } from 'computesdk';

// Auto-detects Vercel from VERCEL_OIDC_TOKEN or VERCEL_TOKEN environment variables
const sandbox = await compute.sandbox.create();

// Execute Node.js command
const result = await sandbox.runCommand('node -e "console.log(\"Hello from Vercel!\")"');
console.log(result.stdout); // "Hello from Vercel!"

// Execute Python command
const pythonResult = await sandbox.runCommand('python -c "print(\"Hello from Python!\")"');
console.log(pythonResult.stdout); // "Hello from Python!"

await sandbox.destroy();

Direct Mode

For direct SDK usage without the gateway:

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

const compute = vercel({
  token: process.env.VERCEL_TOKEN,
  teamId: process.env.VERCEL_TEAM_ID, 
  projectId: process.env.VERCEL_PROJECT_ID,
  runtime: 'python',
  timeout: 600000 // 10 minutes
});

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

const result = await sandbox.runCommand('node -e "console.log(\"Hello from Vercel!\")"');
console.log(result.stdout);

await sandbox.destroy();

Configuration

Environment Variables

# Method 1: OIDC Token (Recommended)
export VERCEL_OIDC_TOKEN=your_oidc_token_here

# Method 2: Traditional
export VERCEL_TOKEN=your_vercel_token_here
export VERCEL_TEAM_ID=your_team_id_here
export VERCEL_PROJECT_ID=your_project_id_here

Configuration Options

interface VercelConfig {
  /** Vercel API token - if not provided, will use VERCEL_TOKEN env var */
  token?: string;
  /** Vercel team ID - if not provided, will use VERCEL_TEAM_ID env var */
  teamId?: string;
  /** Vercel project ID - if not provided, will use VERCEL_PROJECT_ID env var */
  projectId?: string;
  /** Default runtime environment */
  runtime?: 'node' | 'python';
  /** Execution timeout in milliseconds */
  timeout?: number;
}

Features

• ✅ Code Execution - Node.js 22 and Python 3.13 runtime support
• ✅ Command Execution - Run shell commands in sandbox
• ✅ Filesystem Operations - Full file system access via shell commands
• ✅ Auto Runtime Detection - Automatically detects Python vs Node.js
• ✅ Long-running Tasks - Up to 45 minutes execution time
• ✅ Global Infrastructure - Runs on Vercel's global network
• ❌ Interactive Terminals - Not supported by Vercel Sandbox
• ❌ Sandbox Reconnection - Sandboxes are ephemeral (single-use)

API Reference

Code Execution

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

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

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

Command Execution

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

// Install packages (Node.js)
const result = await sandbox.runCommand('npm', ['install', 'lodash']);

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

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

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');

Sandbox Management

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

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

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

// Note: Vercel doesn't support listing all sandboxes
// Each sandbox is ephemeral and single-use

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

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

try {
  const compute = vercel({ 
    token: process.env.VERCEL_TOKEN,
    teamId: process.env.VERCEL_TEAM_ID,
    projectId: process.env.VERCEL_PROJECT_ID
  });
  const sandbox = await compute.sandbox.create();
  
  const result = await sandbox.runCommand('invalid code');
} catch (error) {
  if (error.message.includes('Missing Vercel authentication')) {
    console.error('Set VERCEL_OIDC_TOKEN or VERCEL_TOKEN environment variables');
  } else if (error.message.includes('authentication failed')) {
    console.error('Check your Vercel credentials');
  } else if (error.message.includes('team/project configuration failed')) {
    console.error('Check your VERCEL_TEAM_ID and VERCEL_PROJECT_ID');
  } else if (error.message.includes('Syntax error')) {
    console.error('Code has syntax errors');
  }
}

Examples

Node.js Web Server Simulation

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

const compute = vercel({ runtime: 'node' });
const sandbox = await compute.sandbox.create();

const result = await sandbox.runCommand(`node - <<'JS'
const http = require('http');
const url = require('url');

// Simulate API endpoints
const routes = {
  '/api/users': () => ({
    users: [
      { id: 1, name: 'Alice', role: 'Developer' },
      { id: 2, name: 'Bob', role: 'Designer' }
    ]
  }),
  '/api/health': () => ({ 
    status: 'healthy', 
    timestamp: new Date().toISOString() 
  })
};

// Process request
const path = '/api/users';
const response = routes[path] ? routes[path]() : { error: 'Not found' };

console.log('Response:', JSON.stringify(response, null, 2));
JS`);

console.log(result.stdout);
await sandbox.destroy();

Python Data Processing

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

const compute = vercel({ runtime: 'python' });
const sandbox = await compute.sandbox.create();

const result = await sandbox.runCommand(`python - <<'PY'
import json
import statistics
from collections import Counter

# Sample data
sales_data = [
    {"product": "laptop", "quantity": 5, "price": 999},
    {"product": "mouse", "quantity": 20, "price": 25},
    {"product": "keyboard", "quantity": 15, "price": 75},
    {"product": "laptop", "quantity": 3, "price": 999},
    {"product": "mouse", "quantity": 10, "price": 25}
]

# Aggregate sales
product_sales = {}
for sale in sales_data:
    product = sale["product"]
    revenue = sale["quantity"] * sale["price"]
    product_sales[product] = product_sales.get(product, 0) + revenue

# Calculate statistics
revenues = list(product_sales.values())
total_revenue = sum(revenues)
avg_revenue = statistics.mean(revenues)

print(f"Total Revenue: ${total_revenue}")
print(f"Average Revenue per Product: ${avg_revenue:.2f}")
print("\\nRevenue by Product:")
for product, revenue in sorted(product_sales.items(), key=lambda x: x[1], reverse=True):
    print(f"  {product}: ${revenue}")
PY`);

console.log(result.stdout);
await sandbox.destroy();

Filesystem Operations Pipeline

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

const compute = vercel({ runtime: 'python' });
const sandbox = await compute.sandbox.create();

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

// Create configuration file
const config = {
  project_name: "Vercel Data Pipeline",
  version: "1.0.0",
  settings: {
    input_format: "json",
    output_format: "csv",
    debug: true
  }
};

await sandbox.filesystem.writeFile(
  '/tmp/project/config.json', 
  JSON.stringify(config, null, 2)
);

// Create sample data
const sampleData = [
  { id: 1, name: "Alice", department: "Engineering", salary: 95000 },
  { id: 2, name: "Bob", department: "Marketing", salary: 75000 },
  { id: 3, name: "Charlie", department: "Engineering", salary: 105000 },
  { id: 4, name: "Diana", department: "Sales", salary: 85000 }
];

await sandbox.filesystem.writeFile(
  '/tmp/project/data/employees.json',
  JSON.stringify(sampleData, null, 2)
);

// Process data
const result = await sandbox.runCommand(`python - <<'PY'
import json
import csv
from collections import defaultdict

# Read configuration
with open('/tmp/project/config.json', 'r') as f:
    config = json.load(f)

print(f"Running {config['project_name']} v{config['version']}")

# Read employee data
with open('/tmp/project/data/employees.json', 'r') as f:
    employees = json.load(f)

# Process data - calculate department statistics
dept_stats = defaultdict(list)
for emp in employees:
    dept_stats[emp['department']].append(emp['salary'])

# Calculate averages
results = []
for dept, salaries in dept_stats.items():
    avg_salary = sum(salaries) / len(salaries)
    results.append({
        'department': dept,
        'employee_count': len(salaries),
        'average_salary': round(avg_salary, 2),
        'total_salary': sum(salaries)
    })

# Sort by average salary
results.sort(key=lambda x: x['average_salary'], reverse=True)

# Write results as JSON
with open('/tmp/project/output/department_stats.json', 'w') as f:
    json.dump(results, f, indent=2)

# Write results as CSV
with open('/tmp/project/output/department_stats.csv', 'w', newline='') as f:
    writer = csv.DictWriter(f, fieldnames=['department', 'employee_count', 'average_salary', 'total_salary'])
    writer.writeheader()
    writer.writerows(results)

print("Processing complete!")
print(f"Generated {len(results)} department statistics")

# Print summary
for result in results:
    print(f"{result['department']}: {result['employee_count']} employees, avg salary ${result['average_salary']}")
PY`);

console.log('Execution Output:', result.stdout);

// Read and display results
const jsonResults = await sandbox.filesystem.readFile('/tmp/project/output/department_stats.json');
const csvResults = await sandbox.filesystem.readFile('/tmp/project/output/department_stats.csv');

console.log('JSON Results:', jsonResults);
console.log('CSV Results:', csvResults);

// List all generated files
const outputFiles = await sandbox.filesystem.readdir('/tmp/project/output');
console.log('Generated files:');
outputFiles.forEach(file => {
  console.log(`  ${file.name} (${file.size} bytes)`);
});

await sandbox.destroy();

Package Installation and Usage

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

const compute = vercel({ runtime: 'node' });
const sandbox = await compute.sandbox.create();

// Install lodash
const installResult = await sandbox.runCommand('npm', ['install', 'lodash']);
console.log('Install result:', installResult.stdout);

// Use lodash in code
const result = await sandbox.runCommand(`node - <<'JS'
const _ = require('lodash');

const data = [
  { name: 'Alice', age: 25, city: 'New York' },
  { name: 'Bob', age: 30, city: 'San Francisco' },
  { name: 'Charlie', age: 35, city: 'Chicago' }
];

// Group by city
const grouped = _.groupBy(data, 'city');
console.log('Grouped by city:', JSON.stringify(grouped, null, 2));

// Calculate average age
const avgAge = _.meanBy(data, 'age');
console.log('Average age:', avgAge);

// Find oldest person
const oldest = _.maxBy(data, 'age');
console.log('Oldest person:', oldest.name);
JS`);

console.log(result.stdout);
await sandbox.destroy();

Best Practices

1. Authentication: Use OIDC token method when possible for simpler setup
2. Resource Management: Destroy sandboxes when done (they're ephemeral anyway)
3. Error Handling: Use try-catch blocks for robust error handling
4. Timeouts: Set appropriate timeouts for long-running tasks (up to 45 minutes)
5. File Organization: Use the filesystem API to organize project files
6. Package Installation: Install packages at runtime as needed

Limitations

• Ephemeral Sandboxes: Each sandbox is single-use and cannot be reconnected to
• No Sandbox Listing: Vercel doesn't support listing all sandboxes
• No Interactive Terminals: Terminal operations are not supported
• Memory Limits: Subject to Vercel sandbox memory constraints (2048 MB per vCPU)
• Execution Time: Maximum 45 minutes execution time
• Network Access: Limited outbound network access

Support

• Vercel Documentation
• ComputeSDK Issues
• Vercel Support

License

MIT