Skip to main content
skill CX Codex

minecraft-world-generation — Claude Code skill for Minecraft worldgen

minecraft-world-generation is a Claude Code skill for building and debugging Minecraft world generation in datapacks, NeoForge, or Fabric. It activates for biome, dimension, structure, and worldgen registration tasks.

MIT 269 installs ★ 211 updated 2026-10-11

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add Jahrome907/minecraft-agent-skills --skill minecraft-world-generation -a claude-code
Open on GitHub

What minecraft-world-generation does

The skill helps create and troubleshoot world-generation data and its registration, covering biomes, dimensions, features, and structures. It handles datapacks and mod workflows for NeoForge or Fabric, not gameplay systems unrelated to world generation.

For new Minecraft 26.x work, use the schema from the exact target version or its generated data, along with Java 25. Existing 1.21.x projects keep Java 21 and their matching schema unless the task includes an upgrade. In 26.3, feature files use worldgen/feature/; versions through 26.2 use worldgen/configured_feature/.

The data graph starts with a feature and its placed feature, then connects them to a biome or biome modifier. Structures can also depend on a template pool and processor list; dimensions reference a dimension type and noise settings. Non-worldgen datapack tasks belong to minecraft-datapack, while non-worldgen mod gameplay code belongs to minecraft-modding.

What it helps with

  • create a biome or dimension for a Minecraft version
  • place features in a biome at the intended decoration step
  • add features or spawns through a biome modifier
  • connect a structure to its supporting worldgen data
  • debug worldgen schema mismatches between versions

Example prompts for the agent

  • Create a biome for my Minecraft 26.x datapack.
  • Find why a placed feature is missing from this NeoForge biome.
  • Update this project's worldgen data from 1.21.x to its target version.

How to install minecraft-world-generation

1Skills CLI (Vercel)

npx skills add Jahrome907/minecraft-agent-skills --skill minecraft-world-generation -a claude-code

2Manually: put the files in place

~/.claude/skills/minecraft-world-generation/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/minecraft-world-generation/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about minecraft-world-generation

How do I install minecraft-world-generation in Claude Code?

Run npx skills add Jahrome907/minecraft-agent-skills --skill minecraft-world-generation -a claude-code. The skill file is at ~/.claude/skills/minecraft-world-generation/SKILL.md.

What does minecraft-world-generation do?

It helps build and debug world-generation data and registration for datapacks, NeoForge, and Fabric. Its scope includes biomes, dimensions, features, and structures.

Which Java version and schema should I use?

For new Minecraft 26.x work, use Java 25 and the schema for the exact target version. Existing 1.21.x projects use Java 21 and their matching schema unless the task calls for an upgrade.

Source text

SKILL.md · MIT
---
name: minecraft-world-generation
description: "Create and debug Minecraft 26.x and legacy 1.21.x world generation for datapacks, NeoForge, or Fabric, including biomes, dimensions, features, structures, and biome modifiers. Use for worldgen data or registration, not general gameplay systems."
---

# Minecraft World Generation

Use this skill for biome, dimension, feature, or structure data and their
registration. Use `minecraft-datapack` for non-worldgen data and
`minecraft-modding` for non-worldgen gameplay code.

## Routing Boundaries

- `Use when`: the task changes worldgen data, registration, or injection.
- `Do not use when`: the task is non-worldgen datapack work (`minecraft-datapack`).
- `Do not use when`: the task is non-worldgen mod systems (`minecraft-modding`).

## Choose the delivery path

| Approach | Best When | Platform |
|----------|-----------|----------|
| Datapack JSON | Change data supplied by a pack | Vanilla, any server |
| **Mod + Datagen** | Registering new biomes/dimensions, code-driven | NeoForge / Fabric |
| **Biome Modifier (NeoForge)** | Adding features/spawns to existing biomes | NeoForge |
| **BiomeModification API (Fabric)** | Adding features/spawns to existing biomes | Fabric |

Worldgen registries are datapack registries: their files load at world load and
their registry path determines the data path. Read the [NeoForge registry
guide](https://docs.neoforged.net/docs/concepts/registries/) before choosing a
mod-specific registry path.

## Version boundary

Use Minecraft 26.x for new work. Use Java 25 and start
each JSON schema from the exact target's vanilla data or generated output. Do
not copy a 1.21 shape into a 26.x pack merely because it parses as JSON.

Preserve an established 1.21.x project on Java 21 and its matching schema unless
the task explicitly includes an upgrade. Keep examples matched to the project version.

The [26.1 migration primer](https://docs.neoforged.net/primer/docs/26.1/)
removes `minecraft:random_patch` and `minecraft:no_bonemeal_flower`. It replaces
the random-patch pattern with a separate `minecraft:simple_block` configured
feature and placements for count, random offset, and block-predicate filtering.
Inspect the relevant primer section before migrating code or data.

Read [legacy 1.21 JSON patterns](references/legacy-1.21-worldgen-json.md) only
when the project targets that version. Those examples are not
release artifacts for 26.x.

---

## Data layout and reference graph

26.3 uses `worldgen/feature/`; through 26.2 this was `worldgen/configured_feature/`.
In 26.3, feature fields are at the root, block states use `id` / `properties`,
and biome spawn settings belong in environment attributes. See the
[26.3 migration primer](https://docs.neoforged.net/primer/docs/26.3/).

```
data/<namespace>/
├── worldgen/
│   ├── biome/
│   │   └── my_biome.json
│   ├── feature/
│   │   └── my_ore.json
│   ├── placed_feature/
│   │   └── my_ore_placed.json
│   ├── noise_settings/
│   │   └── my_dimension_noise.json
│   ├── structure/
│   │   └── my_structure.json
│   ├── structure_set/
│   │   └── my_structures.json
│   ├── processor_list/
│   │   └── my_processors.json
│   ├── template_pool/
│   │   └── my_pool.json
│   └── carver/
│       └── my_carver.json
├── dimension/
│   └── my_dimension.json
├── dimension_type/
│   └── my_type.json
├── tags/
│   └── worldgen/
│       └── biome/
│           └── is_forest.json
└── neoforge/
    └── biome_modifier/      (NeoForge mod only)
        └── add_ores.json
```

Build and review the graph from its leaves upward:

1. Define a feature, then its placed feature.
2. Reference placed features from a biome or biome modifier at the intended
   decoration step.
3. Define a structure, then its structure set; a jigsaw structure also needs a
   template pool, processor list, and structure template.
4. Define a dimension type and noise settings before a dimension that references
   them.

Use fully qualified identifiers across namespaces. An external `minecraft:` or
dependency reference is valid when that dependency supplies the registry entry;
do not create a local copy merely to satisfy static checking. If the same pack
contains that external namespace and registry directory, treat it as local and
verify the target exists.

---

## Biomes and dimensions

For 26.x biome and dimension data, use the exact target's vanilla data or
datagen output as the schema source. The older `effects` and dimension-type
fields do not model newer environment behavior. The 11 decoration steps still
organize placed features; choose the semantically appropriate step and keep ore
placement in `underground_ores`.

The version-labeled 1.21.5 biome and dimension examples are in
[legacy 1.21 JSON patterns](references/legacy-1.21-worldgen-json.md).

## 26.3 feature pattern

This berry-bush example follows the 26.3 vanilla data. Keep older packs on
their original schema; do not port them by changing metadata alone.

At `data/<namespace>/worldgen/feature/my_plant.json`:

```json
{
  "type": "minecraft:simple_block",
  "to_place": {
    "id": "minecraft:sweet_berry_bush",
    "properties": { "age": "3" }
  }
}
```

At `data/<namespace>/worldgen/placed_feature/my_plant.json`:

```json
{
  "feature": "<namespace>:my_plant",
  "placement": [
    { "type": "minecraft:count", "count": 96 },
    {
      "type": "minecraft:offset",
      "x": { "type": "minecraft:trapezoid", "min": -7, "max": 7, "plateau": 0 },
      "y": { "type": "minecraft:trapezoid", "min": -3, "max": 3, "plateau": 0 },
      "z": { "type": "minecraft:trapezoid", "min": -7, "max": 7, "plateau": 0 }
    },
    {
      "type": "minecraft:block_predicate_filter",
      "predicate": {
        "type": "minecraft:all_of",
        "predicates": [
          { "type": "minecraft:matching_block_tag", "tag": "minecraft:air" },
          { "type": "minecraft:matching_blocks", "blocks": "minecraft:grass_block", "offset": [0, -1, 0] }
        ]
      }
    }
  ]
}
```

---

## NeoForge biome modifiers

Biome modifiers load from
`data/<modid>/neoforge/biome_modifier/<path>.json`. They can target a biome id
or tag and add or remove placed features, among other changes. The current
[Biome Modifiers guide](https://docs.neoforged.net/docs/worldgen/biomemodifier/)
documents their schemas, decoration steps, and datagen.

For `neoforge:add_features`, `features` accepts a placed-feature id, list, or
tag. Vanilla placed features may be referenced in biome JSON or added with a
modifier, but NeoForge cautions against doing both because feature-order cycles
can crash world loading. Prefer a copy under the mod namespace when an injected
vanilla feature would create that risk.

When targeting a biome from an optional dependency, put the target in a biome
tag entry with `required: false`, then use that tag in the modifier. This lets
the pack load when the dependency is absent.

---

## Structures and dimensions

For any current release, derive structure, template-pool, dimension, and
dimension-type JSON from that release's vanilla data or datagen output. Confirm
the reference graph before launching a test world:

- `structure_set` references `structure`.
- Jigsaw `start_pool` references `template_pool`; each single-pool element
  references its structure template and processor list.
- `dimension.type` references `dimension_type`; a noise generator's string
  `settings` references `worldgen/noise_settings`.

For Fabric registration or mod datagen, use the exact loader and API version's
documentation rather than copying 1.21 code into a 26.x project.

The detailed 1.21 structure and dimension examples are in
[legacy 1.21 JSON patterns](references/legacy-1.21-worldgen-json.md).

---

## Development Workflow

1. Create or edit worldgen JSON files in `data/<namespace>/worldgen/` (or equivalent mod resources path).
2. Run the bundled validator to catch JSON and cross-reference errors before loading:
   ```bash
   ./scripts/validate-worldgen-json.sh --root /path/to/datapack-or-mod-resources --feature-registry feature
   # Strict mode treats warnings as failures:
   ./scripts/validate-worldgen-json.sh --root /path/to/datapack-or-mod-resources --feature-registry feature --strict
   ```
   Omit `--feature-registry feature` for packs through 26.2.
3. Fix any reported errors and re-validate until clean. The validator checks:
   - JSON validity for `worldgen/**` and `neoforge/biome_modifier/**`
   - Cross-reference integrity for `placed_feature -> feature` (or legacy `configured_feature`)
   - Cross-reference integrity for `structure_set -> structure` and biome/biome_modifier feature targets
   - Cross-reference integrity for `jigsaw structure -> start_pool` and `template_pool -> structure template / processor_list`
4. Compare biome and dimension-type JSON against the exact target's vanilla
   registry shape before in-game testing. The helper does not run Mojang codecs:
   valid JSON and local references do not prove that fields such as `carvers`,
   `effects`, or dimension settings match that release's schema.
5. In-game biome and structure testing:
   ```mcfunction
   /locate structure <namespace>:my_structure
   /locate biome <namespace>:my_biome
   /place feature <namespace>:my_ore
   ```
   `place feature` takes a feature ID (`configured_feature` through 26.2),
   not its placed-feature wrapper.
6. For dimension testing, use `/execute in` (dimension must exist at world load, not added via `/reload`):
   ```mcfunction
   execute in <namespace>:my_dimension run tp @s 0 100 0
   ```
7. Check `latest.log` for worldgen errors (missing biome references, malformed noise settings).
8. Note: `/reload` refreshes datapack JSON but does **not** re-generate already-generated chunks. Test new worldgen in a fresh world or newly generated chunks. For existing test worlds, use a disposable copy and a purpose-built chunk reset/regeneration workflow; `/fill` only replaces blocks and is not a substitute for world generation.

---

## References

- Minecraft Wiki — World generation: https://minecraft.wiki/w/Custom_world_generation
- Minecraft Wiki — Biome: https://minecraft.wiki/w/Biome/JSON_format
- Minecraft Wiki — Features: https://minecraft.wiki/w/World_generation/Configured_feature
- NeoForge Biome Modifiers: https://docs.neoforged.net/docs/worldgen/biomemodifier/
- Fabric BiomeModifications: https://wiki.fabricmc.net/tutorial:biomemodification
- misode's data pack generator (worldgen UI): https://misode.github.io/worldgen/

© Jahrome907 · MIT · Jahrome907/minecraft-agent-skills

Similar skills

All skills for Game dev and 3D: Unity, Unreal, Godot, Blender →

From the same set13

The whole set →
minecraft-testing minecraft-testing writes and runs automated tests for Minecraft mods and plugins using JUnit, MockBukkit, NeoForge GameTests, and Fabric GameTests. CX Codex +1 386installs minecraft-resource-pack minecraft-resource-pack is a skill that generates and troubleshoots Minecraft 1.21.x and 26.x resource packs, including block models, sound maps, and pack.mcmeta manifests. CX Codex +1 385installs minecraft-datapack The minecraft-datapack skill for Claude Code creates and debugs vanilla Minecraft 1.21.x and 26.x datapacks. It builds folder structures, functions, loot tables, and pack.mcmeta without mods. CX Codex +1 300installs minecraft-server-admin minecraft-server-admin runs the infrastructure side of Minecraft Java servers: Paper, Purpur, Folia, Velocity, Fabric, NeoForge. Useful when TPS drops or a proxy blocks logins. CX Codex +1 279installs minecraft-plugin-dev minecraft-plugin-dev helps build, change, and debug Minecraft server plugins for Paper and Bukkit-derived servers. Useful for JavaPlugin APIs, events, commands, and configuration. CX Codex +1 267installs minecraft-multiloader minecraft-multiloader is a Claude Code skill that builds and maintains Architectury Minecraft mods for 26.x and 1.21.x across NeoForge and Fabric. It runs when both loaders need release artifacts. CX Codex +1 247installs minecraft-commands-scripting minecraft-commands-scripting writes and debugs Minecraft Java 26.x and 1.21.x commands, execute chains, scoreboards, and RCON scripts. Targets command logic directly without generating full datapack folder structures. CX Codex +1 241installs minecraft-ci-release minecraft-ci-release sets up CI and release plumbing for Minecraft mods and Paper plugins: workflow files, artifacts, versions, tags, GitHub Releases. CX Codex +1 238installs

Project sponsors

Companies that keep this analytics open