Security is crucial when working with MCP servers and LLM agents. This guide covers best practices for protecting your applications, data, and infrastructure.
Important: MCP servers can have powerful capabilities including file system access, network requests, and code execution. Always follow security best practices to protect your systems.
API Key Management
Environment Variables
Never hardcode API keys in your source code. Use environment variables:
import { config } from 'dotenv'
// Load environment variables from .env file
config()
// Access API keys securely
const openaiKey = process.env.OPENAI_API_KEY
const anthropicKey = process.env.ANTHROPIC_API_KEY
if (!openaiKey) {
throw new Error('OPENAI_API_KEY environment variable is required')
}
.env File Security
Create a secure .env file:
# LLM Provider Keys
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GROQ_API_KEY=gsk_...
# MCP Server Configuration
FILESYSTEM_ROOT=/safe/workspace
DATABASE_URL=postgresql://user:pass@localhost/db
# Optional: Security settings
MCP_TIMEOUT=30
MAX_TOOL_CALLS=10
ALLOWED_DOMAINS=example.com,api.service.com
Never commit .env files: Add .env to your .gitignore file to prevent accidentally committing API keys to version control.
Secrets Management
For production environments, use proper secrets management:
AWS Secrets Manager
Azure Key Vault
HashiCorp Vault
import { SecretsManagerClient, GetSecretValueCommand } from '@aws-sdk/client-secrets-manager'
async function getSecret(secretName: string, regionName = 'us-east-1') {
const client = new SecretsManagerClient({ region: regionName })
try {
const response = await client.send(
new GetSecretValueCommand({ SecretId: secretName })
)
return response.SecretString
} catch (error) {
throw error
}
}
// Usage
const openaiKey = await getSecret('prod/mcp-use/openai-key')
MCP Server Security
Filesystem Server Security
When using filesystem servers, restrict access to safe directories:
secure_filesystem_config.json
{
"mcpServers": {
"filesystem": {
"command": "mcp-server-filesystem",
"args": [
"/workspace/safe-directory",
"--readonly",
"--max-file-size", "10MB",
"--allowed-extensions", ".txt,.md,.json,.py"
],
"env": {
"FILESYSTEM_READONLY": "true",
"MAX_FILE_SIZE": "10485760"
}
}
}
}
Network Access Restrictions
Limit network access for web-based MCP servers:
secure_network_config.json
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": ["@playwright/mcp@latest"],
"env": {
"PLAYWRIGHT_HEADLESS": "true",
"ALLOWED_DOMAINS": "example.com,api.trusted-service.com",
"BLOCK_PRIVATE_IPS": "true",
"DISABLE_JAVASCRIPT": "false",
"TIMEOUT": "30000"
}
}
}
}
Database Security
Secure database connections with proper credentials and restrictions:
secure_database_config.json
{
"mcpServers": {
"postgres": {
"command": "mcp-server-postgres",
"env": {
"DATABASE_URL": "${DATABASE_URL}",
"CONNECTION_TIMEOUT": "30",
"MAX_CONNECTIONS": "5",
"READONLY_MODE": "true",
"ALLOWED_SCHEMAS": "public,reporting",
"BLOCKED_TABLES": "users,passwords,secrets"
}
}
}
}
Agent Security Configuration
Limit which tools the agent can use:
Validate user inputs before processing:
Rate Limiting
Implement rate limiting to prevent abuse:
Logging and Monitoring
Security Logging
Implement comprehensive security logging:
Monitoring Dashboard
Create monitoring for security events:
Production Deployment Security
Container Security
Use secure container configurations:
FROM node:18-alpine
# Create non-root user
RUN addgroup -S mcpuser && adduser -S mcpuser -G mcpuser
# Install security updates
RUN apk update && apk upgrade && \
apk add --no-cache ca-certificates && \
rm -rf /var/cache/apk/*
# Set working directory
WORKDIR /app
# Copy package files and install dependencies
COPY package*.json ./
RUN npm ci --only=production
# Copy application code
COPY . .
# Set ownership and permissions
RUN chown -R mcpuser:mcpuser /app
USER mcpuser
# Expose port
EXPOSE 8000
# Run application
CMD ["node", "main.js"]
Network Security
Configure network policies and firewalls:
kubernetes_network_policy.yaml
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: mcp-use-policy
spec:
podSelector:
matchLabels:
app: mcp-use
policyTypes:
- Ingress
- Egress
ingress:
- from:
- podSelector:
matchLabels:
app: frontend
ports:
- protocol: TCP
port: 8000
egress:
- to: []
ports:
- protocol: TCP
port: 443 # HTTPS only
- to:
- namespaceSelector:
matchLabels:
name: database
ports:
- protocol: TCP
port: 5432
Security Checklist
Common Security Vulnerabilities
Path Traversal Prevention
Command Injection Prevention
Next Steps
Security is an ongoing process. Regularly review and update your security practices, monitor for new vulnerabilities, and keep all dependencies up to date.
