[rom1504bot]

This automated PR sets up the relevant boilerplate for Minecraft pc version 26.2. Fixes #1218.

Related:

• Issue: Support Minecraft PC 26.2 #1197
• Protocol Version: 776

• You can help contribute to this PR by opening a PR against this pc_26_2 branch instead of master.

[extremeheat]

@copilot please re-add the changes made in #1216 to this updated PR for 26.2 and also correctly update the YAML files

[Copilot]

@copilot please re-add the changes made in #1216 to this updated PR for 26.2 and also correctly update the YAML files

Addressed in 4dd8762. The 26.2 login field fixes from #1216 are re-applied in data/pc/latest/proto.yml, and data/pc/26.2/protocol.json was regenerated from that YAML source.

[perceiveing]

Heads up — the generated data/pc/26.2/protocol.json in this PR is missing two wire changes that are present in the 26.2 server jar. Both look like the extractor carrying 26.1 definitions forward.

I verified this by decompiling the 26.1 and 26.2 server jars (26.x ships deobfuscated, so no mappings needed) and diffing the whole net/minecraft/network/ tree. Fifteen files differ; eleven are refactors with no wire effect. Four are real. This PR has two of them (packet_success / sessionId and packet_login / onlineMode — both correct, and I confirmed onlineMode is placed correctly between commonPlayerSpawnInfo and enforcesSecureChat). The other two are missing.

1. packet_teams — Parameters is fully reordered

ClientboundSetPlayerTeamPacket.Parameters became a record with a StreamCodec in 26.2, and the field order changed:

#	26.1	26.2
1	displayName	displayName
2	options (byte)	playerPrefix
3	nametagVisibility	playerSuffix
4	collisionRule	nameTagVisibility
5	color — varint ChatFormatting ordinal	collisionRule
6	playerPrefix	color — Optional<TeamColor>
7	playerSuffix	options (byte)

26.2 source:

public record Parameters(
   Component displayName, Component playerPrefix, Component playerSuffix,
   Team.Visibility nameTagVisibility, Team.CollisionRule collisionRule,
   Optional<TeamColor> color, @PlayerTeam.OptionFlags byte options
) {
   public static final StreamCodec<RegistryFriendlyByteBuf, Parameters> STREAM_CODEC = StreamCodec.composite(
      ComponentSerialization.TRUSTED_STREAM_CODEC, Parameters::displayName,
      ComponentSerialization.TRUSTED_STREAM_CODEC, Parameters::playerPrefix,
      ComponentSerialization.TRUSTED_STREAM_CODEC, Parameters::playerSuffix,
      Team.Visibility.STREAM_CODEC,               Parameters::nameTagVisibility,
      Team.CollisionRule.STREAM_CODEC,            Parameters::collisionRule,
      ByteBufCodecs.optional(TeamColor.STREAM_CODEC), Parameters::color,
      ByteBufCodecs.BYTE,                         Parameters::options,
      Parameters::new
   );
}

Two things beyond the reorder:

• options moves from slot 2 to slot 7, and prefix/suffix move from the tail to slots 2–3.
• color is now Optional<TeamColor> — a bool prefix followed by the varint if present. The domain also narrows: 26.1 wrote a ChatFormatting ordinal (0–21, with 21 = RESET meaning "no color"), whereas TeamColor is 0–15 and absence is carried by the Optional instead. TeamColor.STREAM_CODEC is ByteBufCodecs.idMapper(BY_ID, v -> v.id), so plain varint.

The current definition in this PR is still the 26.1 layout (name, flags, nameTagVisibility, collisionRule, formatting, prefix, suffix), so it will desync on the first team packet a 26.2 server sends. Worth noting this is the failure mode that parses "successfully" into wrong fields rather than throwing, so it may not show up in a connect-and-spawn smoke test.

2. spectate_entity replaced by spectator_action

- .addPacket(GamePacketTypes.SERVERBOUND_SPECTATE_ENTITY, ServerboundSpectateEntityPacket.STREAM_CODEC)
+ .addPacket(GamePacketTypes.SERVERBOUND_SPECTATOR_ACTION, ServerboundSpectatorActionPacket.STREAM_CODEC)

public record ServerboundSpectatorActionPacket(OptionalInt spectateEntityId) {
   public static final StreamCodec<ByteBuf, ServerboundSpectatorActionPacket> STREAM_CODEC =
      StreamCodec.composite(ByteBufCodecs.OPTIONAL_VAR_INT, ..., ServerboundSpectatorActionPacket::new);
}

ByteBufCodecs.VAR_INT → ByteBufCodecs.OPTIONAL_VAR_INT, and the packet is renamed. It occupies the same slot in GameProtocols, so packet IDs do not shift. This PR still has packet_spectate_entity: { entityId: varint }. Serverbound and rarely used, so low practical impact — but it is wrong.

Separately: the dataPaths.json aliases look unsafe for this version

This PR points blocks, items, entities, particles, sounds, attributes, biomes and recipes at pc/26.1. 26.2 is not registry-compatible with 26.1 — I ran the vanilla --reports generator on both jars and diffed all 95 registries:

registry	26.1 → 26.2	protocol IDs shifted
item	1506 → 1537	1480 of 1506
sound_event	1902 → 1968	1674
block	1168 → 1196	170
particle_type	117 → 125	113
block_type	263 → 265	104
game_event	60 → 61	51
attribute	35 → 40	all 35
data_component_type	110 → 111	32
entity_type	157 → 158	27
block_entity_type	49 → 49	23

26.2 adds a cinnabar/sulfur block family, a sulfur_cube entity, geyser/noxious-gas particles and five new attributes, all inserted mid-registry, which renumbers most of what follows. block_entity_type also removes minecraft:bed.

Block states matter most for chunk decoding: 29,873 → 32,366, and 5,186 state IDs that are valid in both versions name a different block. The first divergence is at 24687:

id 24687:  26.1 = calcite        ->  26.2 = sulfur
id 24688:  26.1 = tinted_glass   ->  26.2 = potent_sulfur
id 24689:  26.1 = powder_snow    ->  26.2 = potent_sulfur

Chunks would decode without error and report plausible-looking wrong blocks. Happy to open a PR against pc_26_2 with extracted 26.2 tables and the two protocol fixes above if that's useful.

One small correction while I'm here, in case it saves someone time: the global block-state palette for 26.2 needs 15 bits, not 17 — ceil(log2(32366)) = 15, same as 26.1. I mention it because PrismarineJS/prismarine-chunk#331 relaxes prismarine-chunk's bitsPerBlock > 16 guard on the stated basis that 26.x has "more than 65536 block states". It has 32,366, and with correct data maxBitsPerBlock computes to 15, so that guard should never legitimately fire. If it is firing in practice, the stale registry data above is a more likely cause than the bound.

Generated with Claude Code

[rom1504]

Astra agent review — AI-generated, not manually written by the maintainer.

Not ready to merge at 68ea7b5: Regenerate26.1 YAML/JSON consistently, restore data CI, then finish26.2 cross-module integration.

The older comments about teams ordering, optional team color, spectator_action, and the main registry aliases are addressed in the current source. I am not repeating those resolved findings. Full cross-module 26.2 integration still needs its normal version-branch checks.

Validation: current head and discussion rechecked; this review uses the source, CI and focused evidence collected for this same head earlier today. No additional full-suite run in this publication pass.

Skills used: prismarine-review reconciled the current code with existing feedback and validation; prismarine-protocol-data-review traced maintained schemas/generators through version selection and generated data.

[rom1504]

Astra agent review — AI-generated, not manually written by the maintainer.

The new frozen 26.1 YAML selection is not synchronized with its generated JSON: current CI reports 26.1 (26.1) / protocol.json is desynced from yaml, please run npm run build (1,922 passing tests, one failure). Please regenerate the affected YAML/JSON pair and rerun validation before landing the 26.2 data.

Skills used: prismarine-review reconciled the current code with existing feedback and validation; prismarine-protocol-data-review traced maintained schemas/generators through version selection and generated data.

[0xmthan]

The 26.1 desync is fixed in #1333 (targets pc_26_2): it merges master and copies master's 26.1 proto.yml (#1278, #1295) into pc/26.1/, then rebuilds. npm test passes locally (2067 passing). CI there needs approval to run.

[Cyverno]

I opened #1339 against pc_26_2. It builds on #1333 and does a real merge of current master. It also fixes three things I found by checking the 26.2 data against the vanilla 26.2 jar and its --reports output:

• the attribute mapper, which was missing the five new 26.2 attributes, so every key decoded as its neighbour
• 1-based sound ids
• loginPacket.json, which was 26.1's registries. It's now regenerated from a vanilla 26.2 server.

I also checked the packet ids, every registry-backed mapper, block states, items, entities and entity metadata, and they match vanilla. The details and the downstream test results (node-minecraft-protocol, mineflayer and a live Paper 26.2 smoke test) are in #1339. With it, the nmp and mineflayer pc26_2 branches pass their 26.2 suites without the debug protodef builds: PrismarineJS/node-minecraft-protocol#1550 and PrismarineJS/mineflayer#4192.

[extremeheat]

@Cyverno I updated this PR with latest master ; can you please recreate your PR only with the changes ?

[Cyverno]

Done: #1344 is against the current pc_26_2 and has only the 26.2 changes, plus a resync of data/pc/26.1/proto.yml with master so the yaml/json sync test passes on this branch. I also dropped the team_color rename, following the updated naming guidance.