Update hytale protocol to current latest by SantioMC · Pull Request #5 · xyzeva/hyproxy

SantioMC · 2026-06-22T21:43:53Z

Based on internal pr: Minehut#1

…ces deployed engine

…ocol from server jar

…y from JWT; gated byte logging

…l-safe language) and clientType decode

…ction, Collection players)

…d validates)

…token; defer player registration

…forward undecodable packets raw

xyzeva

thanks for trying to update hyproxy to latest, while i really appreciate this pr, its riddled with issues and excessive ai usage and unneeded changes

if you are willing to fix all of my review comments, i can take another look at it. but at its current state it is heavily blocked from being merged and i would say you shouldn't use this in production

if you don't want to fix the issues, i plan to update to latest sometime soon myself anyway.

xyzeva · 2026-07-18T18:52:59Z

            Packet packet = packetInfo.deserializeFunction().apply(payload);
            out.add(packet);
+        } catch (Exception e) {
+            log.warn("failed to decode packet id {} - dropping packet", packetId, e);


this is a security issue, we want these to be captured by the netty pipeline which will properly close the connection after logging, or not log at all if the player hasn't authenticated yet to prevent console spam

xyzeva · 2026-07-18T18:53:41Z

+import java.util.Arrays;
+import java.util.Map;
+
+/**


we dont need this large javadoc for a completely internal class describing all of what it does, it simply isn't needed or wanted

xyzeva · 2026-07-18T18:54:31Z

            for (int i = 0; i < length; i++) {
                int oldParamOffset = buf.readerIndex();
-                String key = ProtocolUtil.readVarString(buf, 128);
+                String key = ProtocolUtil.readVarString(buf, 4096000);


do not use string sizes as large as this, even if the original server uses them. use a more appropriate amount for the situation

xyzeva · 2026-07-18T18:54:36Z

            for (int i = 0; i < length; i++) {
                int oldParamOffset = buf.readerIndex();
-                String key = ProtocolUtil.readVarString(buf, 128);
+                String key = ProtocolUtil.readVarString(buf, 4096000);


do not use string sizes as large as this, even if the original server uses them. use a more appropriate amount for the situation

xyzeva · 2026-07-18T18:54:53Z

        if ((nullBits & 0x2) != 0) {
            int offset = varsOffset + motdOffset;
-            Pair<String, Integer> varString = ProtocolUtil.readVarString(buf, offset, 500);
+            Pair<String, Integer> varString = ProtocolUtil.readVarString(buf, offset, 4096000);


do not use string sizes as large as this, even if the original server uses them. use a more appropriate amount for the situation

xyzeva · 2026-07-18T18:59:22Z

+
+        JWTVerifier.IdentityTokenClaims claims = connection.getProxy().getJwtVerifier().validateIdentityToken(identityToken);
+        if (claims == null) {
+            log.warn("rejecting Connect: identity token failed validation (validateIdentityToken returned null)");


we do not need a log for this

xyzeva · 2026-07-18T19:00:33Z

            return;
        }

+        connection.getProxy().registerPlayer(player);


why was this moved? very intentional decision to not do this

xyzeva · 2026-07-18T19:00:51Z

-            return true;
-        }
-
+        player.setUsername(tokenUsername);


this is a huge security issue, why was this check removed?

xyzeva · 2026-07-18T19:01:59Z

    public void unregisterPlayer(HyProxyPlayer player) {
        if (this.getPlayerByProfileId(player.getProfileId(), true) == null) {
-            throw new IllegalArgumentException("player profile id " + player.getProfileId() + " not registered");
+            return;


this is an intentional decision to return an error here

xyzeva · 2026-07-18T19:02:26Z

+
        this.playersByProfileId.remove(player.getProfileId());
-        this.playersByUsername.remove(player.getUsername().toLowerCase(Locale.ROOT));
+        if (player.getUsername() != null) {


i dont get why this check is needed, but i guess it might be good to have it?

Players intermittently disconnect when the Hytale proxy runs behind Cloudflare Spectrum. Spectrum does not fragment UDP and drops any datagram too large to forward. The client-facing QUIC codec advertised the quiche default max_udp_payload_size (65527) and ran DPLPMTUD (discoverPmtu=true), so datagrams grew past the Spectrum-forwardable size and were blackholed, stalling sessions until the 60s idle timeout dropped the player. Cap the datagram size in both directions and disable PMTU probing, both driven by new config keys so the value can be tuned per deployment without a rebuild: - max-udp-payload-size (default 1200, QUIC's universal floor) sets both maxRecvUdpPayloadSize (advertised to the client, caps client->proxy, the direction Spectrum blackholes) and maxSendUdpPayloadSize (caps proxy->client) - discover-pmtu (default false) drives discoverPmtu(...) Behind Spectrum start at 1200 and raise toward 1350 once stability is confirmed. For direct (non-Spectrum) UDP exposure set discover-pmtu=true.

Addresses the review on xyzeva#5. Security: - PacketDecoder: stop swallowing deserialization failures. Let them propagate so the netty pipeline closes the connection instead of letting a client spam malformed packets while staying connected. Protocol string limits (were 4096000): - FormattedMessage: rawText 4096, messageId 256, param keys 256, color 256, link 4096 - ServerInfo: serverName 256, motd 4096 Applied to all 8 sites, not only the 4 flagged, to stay consistent. Noise removal: - Drop javadoc from internal classes (PlayerSkin, InsecurePlayerOptions, RequestInsecurePlayerOptions) - Drop `final` on locals; it is not used elsewhere in the codebase - Drop verbose explainer comments in Connect and InboundInitialPacketHandler - Drop log.warn on rejected pre-auth connections (console spam vector) - Lowercase internal disconnect messages, matching the existing "invalid referral data" convention Intentionally unchanged: the username now comes from the verified access token rather than a Connect field, because the 0.5.5 Connect packet no longer carries a username. registerPlayer moving after auth, the unregisterPlayer early return and the username null guard all follow from that. Restoring the old equality check would compare against null and reject every login; the UUID cross-check against the token is retained.

alepaez and others added 17 commits June 16, 2026 17:49

docs(MHG-1132): start protocol RE notes; confirm fork base

8e2a394

docs(MHG-1132): record container build recipe; baseline build reprodu…

6f1c6d8

…ces deployed engine

docs(MHG-1132): document current Hytale 0.5.5 handshake/transfer prot…

34c25e9

…ocol from server jar

fix(MHG-1132): rewrite Connect to Hytale 0.5.5 layout; source identit…

526d3f2

…y from JWT; gated byte logging

fix(MHG-1132): harden Connect serialize (fixed 20B ASCII version, nul…

b6f1340

…l-safe language) and clientType decode

fix(MHG-1132): adapt backend-plugin to Hytale 0.5.5 API (ChannelConne…

0283da9

…ction, Collection players)

fix(MHG-1132): drop protocolCrc gate (varies per client build; backen…

96b3674

…d validates)

fix(MHG-1132): source uuid from identity token, username from access …

f8f668c

…token; defer player registration

feat(MHG-1132): implement proxy->backend insecure-options handshake; …

7b07003

…forward undecodable packets raw

fix(MHG-1132): backend plugin derives backend name from SERVER_ID env

3c9f8fb

Merge branch 'main' into alexandresequeira/mhg-1132-update-protocol

b44e941

misc: cleanup & rebase properly

d667675

misc: remove ai documentation

66268f0

feat: log when failing to validate

09e1d25

feat: fix NPE & skins

91e0bbc

fix: dont forward failed packets & formattedmessage

dd4b4cb

misc: more protocol corrections

1989538

xyzeva requested changes Jul 18, 2026

View reviewed changes

alepaez added 2 commits July 21, 2026 14:39

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update hytale protocol to current latest - #5

Update hytale protocol to current latest#5
SantioMC wants to merge 19 commits into
xyzeva:mainfrom
Minehut:alexandresequeira/mhg-1132-update-protocol

SantioMC commented Jun 22, 2026

Uh oh!

xyzeva left a comment

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

xyzeva Jul 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

SantioMC commented Jun 22, 2026

Uh oh!

xyzeva left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants