Minecraft Packager

A framework for dynamic data-driven code generation in Minecraft datapacks, powered by Nunjucks templating.

Interested in working with this project or volunteering to help? Get started here!
Contact Me
Have any questions or concerns? Don't hesitate to reach out!

Open Source

Browse the source code, suggest changes, or develop it yourself!

Git Repo

This is a multipurpose templating & linting tool for Minecraft datapacks & resource packs. It can import and merge other pack data and assets, run on command-line or in the browser, and centralize data or configuration options to template into different parts of your project.

Command-line use

The --type (datapack|resourcepack) flag can be appended to any command to declare a specific package type. By default, the packager will execute both.

minecraft-packager init will initialize the current directory as a datapack/resourcepack.

minecraft-packager build will compile the files in the /datapack and /resourcepack directories. Compiled files will be generated in the /_out directory.

minecraft-packager install will install the compiled resourcepack/datapack to the specified Minecraft location with the --mcdir flag, the $MINECRAFT_HOME environment variable, or ~/.minecraft, in that order.


After initialization, the packager will create a default package.json file with some properties that specify the package information:

  "config": {
    "format": "1.17",
    "dependencies": [
    "zipFiles": {
      "resourcepack": "custom-zip-name.resourcepack.zip",
      "datapack": "custom-zip-name.datapack.zip"

This file is also compatible with npm, so the package can be compiled & installed with just npm run dev.

Templating Syntax

.mcfunction and .json files can make use of a modified Nunjucks templating language to generate code:

  • Custom syntax uses << variable >> and <% tag %> instead of the curly braces.
  • Any .yml or .json files nested in the /_data directory are available as a global variable.
    • For example, the info property in /data/blocks/sponge.yml can be referenced as blocks.sponge.info.
  • Special filters can be used for interacting with JSON or NBT data:
    • << variable | dump >> can be used to output any variable as JSON text
    • << info.nbt_requirements | dump_nbt >> will form an NBT string from the nbt_requirements property in /_data/info.yml

Generated Files

A set of files can be "generated" from an array or object in the /_data directory by prefixing its filename with an underscore and adding some front matter to specify the data to use:

collection: config.teams
filename: '<< key | lower >>.json'
	"name": "<< item.name >>",

collection specifies the object or array to iterate through, and filename generates the file name for each item in the collection.

For each generated item, the following variables are made available for templating (in addition to the global data):

  • key: The item key (for objects, this is the property name; for arrays, the item index)
  • item: The value of the current item
  • index: The current iteration number


  • TODO: parse/validate certain command structures, e.g. execute if entity ... run ...
  • TODO: parse/validate declared scoreboard names & uses
  • TODO: correctly merge loot_tables & other specific JSON files with specific implementation
  • TODO: ability to import files from MC source (implement as node module?)
  • TODO: refactor "datapack" / "resourcepack" dirs to "data" and "assets"