Network policy

Gates can’t scan your home network. The launcher owns every network connection. A gate doesn’t open sockets directly, it asks the launcher’s broker, and the broker decides whether the destination is allowed.

What’s allowed

Public addresses on the open internet. A gate can talk to its own backend, fetch assets from CDNs, connect to multiplayer servers hosted online, and so on. This is the normal case. Most gates need network access to function.

What’s blocked

Anything that points back into your own machine or your local network:

- Your computer. Loopback addresses (127.0.0.1, ::1). A gate can’t reach a development server, a database, or any other service running on the same machine.
- Your home network. Private IPv4 ranges (10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) and the IPv6 equivalents. A gate can’t probe your router, your printer, or other devices on your LAN.
- Link-local addresses. 169.254.0.0/16 and fe80::/10. Often used by WSL, Docker, and zero-config services.
- Carrier-grade NAT. 100.64.0.0/10. Used by Tailscale and some VPNs.
- Multicast. 224.0.0.0/4 and IPv6 multicast. Blocks network-discovery attacks.

Note

The broker enforces this on the launcher side, before a socket is opened. The gate never gets a chance to connect to a blocked destination, because there’s no socket for it to use.

Why this matters

Without the broker, a malicious gate could:

- Map your home network and identify services on it.
- Attack your router’s admin interface (which usually trusts requests from the local network).
- Reach development tools running on your machine that don’t expect hostile input.

With the broker, none of this is reachable. The gate sees the open internet only.

See also