network-security-setup

Configure network isolation for Claude Code sandbox with trusted domains to prevent data exfiltration and malicious access.

• Network isolation policies for AI code execution
• Trusted domain whitelisting to block untrusted access
• Secure environment variable management
• Corporate proxy and internal registry support

• Mode 1: Trusted domains (recommended default)
• Mode 2: No network access (max security)
• Mode 3: Custom access with user-defined domains
• Support for corporate proxies and internal registries
• Secure environment variable handling
• Prevents prompt injection and data exfiltration
• Blocks malicious package downloads
• MUST store secrets in .env.local (never in config)

1. 1.Choose appropriate isolation mode (trusted, none, custom)
2. 2.Define trusted domains based on use case and environment
3. 3.Configure corporate proxy if needed
4. 4.Use safe environment variables only (no secrets)
5. 5.Store secrets in .env.local (gitignored)
6. 6.Validate that policies block untrusted access