Connection Guard vs ProxyShield
Two open-source plugins with different defaults. ProxyShield counts data-centre ranges as a reason to refuse; Connection Guard only refuses evidence of a VPN, proxy or Tor exit.
5.4 MB JAR for every platform, from ModrinthOther download options
Connection Guard 0.6.1 against ProxyShield 2.5.1, measured 7–13 October 2026 on the same server with the same players, each as it ships and without API keys. Facts about ProxyShield from its own listing and configuration, checked 13 October 2026. Written by Connection Guard's author: check the benchmark.
Side by side
| Connection Guard 0.6.1 | ProxyShield 2.5.1 | |
|---|---|---|
| Price | Free | Free |
| Source code | Open (MIT) | Open (GPL-3.0) |
| After installing | Blocks right away: its own signed lists, then zowi, Blackbox and IPQuery, no key needed. | Blocks right away: 7 keyless services vote with weights (score 1.5 refuses), plus lists, reverse DNS and heuristics; data-centre addresses are blocked too. |
| How it decides | Its own lists on your server first, then one service at a time until one gives a clear answer. | Adds up a score from 7 keyless services, IP lists, reverse DNS and join heuristics, and blocks hosting ranges by default. |
| Platforms | Paper, Spigot, Folia, BungeeCord, Velocity: one JAR | Paper, Spigot, Folia, BungeeCord, Velocity, Sponge |
| Shared cache for networks | Redis | Disk cache or Redis |
| Telemetry | bStats; anonymous totals and error reports to its optional dashboard (no player data). Documented, each can be switched off. | bStats. |
| Measured in the benchmark · 7–13 October 2026 | ||
| VPN, Tor and proxy addresses blocked | 363 of 364 (better) | 345 of 364 |
| Home and mobile players refused | 2 of 310 (better) | 8 of 310 |
| VPNs over IPv6 blocked | 40 of 40 (better) | 34 of 40 |
| Public proxies blocked | 85 of 86 (better) | 74 of 86 |
| Stress test, 1,000 players in 20 s: decided within (p95) | 452 ms (better) | 7.3 s |
| Stress test, 1,000 players in 20 s: checked | 1000 (better) | 332 |
| First join of a new player (p95) | 288 ms (better) | 1.4 s |
| Same player again (median) | 3 ms | 2 ms (better) |
| Outages with VPN and Tor still refused | 4 of 4 | 4 of 4 |
The better value in a measured row is set in bold. Speed rows use simulated detection services with the same delay for every plugin; detection rows use the real services, each plugin from a server of its own.
Where Connection Guard is different
Fewer real players refused
Of 310 home and mobile players, ProxyShield refused 8 and Connection Guard 2. ProxyShield blocks data-centre ranges by default, and several home connections sit in ranges some services list as hosting.
Measured
Every player checked in the stress test
In the stress test, with 1,000 players joining in 20 seconds, ProxyShield checked 332 and needed 7.3 s at the 95th percentile; Connection Guard checked 1000 within 452 ms.
Measured
More VPNs over IPv6
ProxyShield blocked 34 of 40 VPN servers reached over IPv6, Connection Guard 40.
Measured
One question at a time
ProxyShield asks its services in parallel for every new address. Connection Guard checks its own lists on the server first and then asks one service at a time, so most players cost one lookup or none, and fewer services see their address.
From its configuration
Where ProxyShield is good
- Open source (GPL-3.0), with dry-run and silent modes and Discord alerts.
- Combines several signals (lists, reverse DNS, join heuristics) instead of trusting one service, and can share its cache through Redis.
- Kept refusing VPNs and Tor exits in every outage the benchmark simulated.
Switching from ProxyShield
Connection Guard 0.6.1 reads ProxyShield's settings. Install it next to ProxyShield, then in the console:
/cg migrate preview proxyshield
It shows what it would take over (mode, countries, API keys, your IP, CIDR, UUID and ASN rules) and what works differently. Apply it with the command it prints, remove ProxyShield and restart; the original files are kept as a backup.
Connection Guard vs ProxyShield: questions
Why does ProxyShield refuse some home players?
It blocks hosting and data-centre ranges by default (detection.block-hosting). Some services list consumer ranges as hosting, so a home player can be refused. You can switch it off in its config.
Can I move my ProxyShield settings to Connection Guard?
Yes. Run /cg migrate preview proxyshield in Connection Guard 0.6.1, review what it maps, then apply it on the next restart.
Benchmark runs: detection, speed, outages. Method and every log: mc-antivpn-bench. ProxyShield's author can contest any number there; corrections are re-measured. ProxyShield is a trademark of its author; this page is not affiliated with it. Listing: Modrinth.