Machine recipes in KubeJS

Every HBM machine table is a KubeJS recipe type. In a server script’s ServerEvents.recipes you write event.recipes.hbm.<table>(...), and event.remove / event.replaceInput / event.replaceOutput work on HBM recipes like on any other.

A script recipe is the same thing as a datapack recipe file: the arguments become the row’s JSON, and the table’s own reader checks it. The JSON forms are the ones in the Machine recipes section.

kubejs/server_scripts/hbm_recipes.js

ServerEvents.recipes(event => {
  // output, input, stamp
  event.recipes.hbm.press('minecraft:diamond', 'minecraft:coal_block', 'flat').id('kubejs:press_diamond')
  // counted output, tag input
  event.recipes.hbm.shredder('2x minecraft:gravel', '#minecraft:logs').id('kubejs:shredder_logs')
})

How the arguments are ordered

Each table states its fields (its JSON keys). Golem Core builds the script function from them:

1. Required fields first, in the order the table states them, then optional fields, in stated order.
2. Every field is also a builder method: .duration(40), .power(100). A field whose name has an underscore answers to both spellings: .input_items(...) and .inputItems(...), .solid_fuel_energy(...) and .solidFuelEnergy(...).
3. Generic tables (assembler, chemical plant, PUREX and the others listed below) have one required field, name, so their first argument is the row’s name. Everything else is optional and usually set with builder methods.

So you can pass optional fields positionally after the required ones, or leave them out and use builder methods. Builder methods are easier to read and do not depend on the order.

Leaving out an optional field means the table’s default applies (the same as leaving the key out of a datapack file).

Recipe ids

Give each recipe an id with .id('namespace:path'). Without one KubeJS makes one up.

What a value can be

A typed argument is read the KubeJS way and then written in the form the field needs:

Field holds	You can pass
item input	'minecraft:iron_ingot', '2x minecraft:iron_ingot', '#c:ingots/steel'
item output	'minecraft:gold_ingot', '3x minecraft:gold_ingot'
fluid	Fluid.of('hbm:oxygen', 500), or a bare id 'hbm:oxygen' (one bucket, 1000 mB)
number, text, true/false	a JS number, string or boolean
raw JSON field	a plain object or array, exactly as in a datapack file

Fluid amounts are millibuckets. On 1.20.1 Fabric KubeJS counts fluids in droplets; the adapter converts them, see Troubleshooting.

Many fields take a narrower form than “any ingredient”. The signature list below names it per field:

Form	Accepts	Rejects
ingredient	one item or one tag, with a count	a list of several items (Ingredient.of([a, b]))
item or tag	one item or one tag	a count other than 1
item	one item	a tag, a count other than 1
item stack in	one item with a count	a tag
tag	one tag	an item, a count other than 1
grid slot	an ingredient, or '' for an empty slot
stack	one item with a count	a chance
item id	one item	a count other than 1, a chance
output	one item with a count and an optional chance
fluid stack	a fluid with an amount
fluid id	a fluid; the amount is ignored

Plain objects pass through

A plain JS object (or a JSON element) given for a field is written unchanged as that field’s JSON. Use this for HBM’s special forms, which have no KubeJS type:

// HBM's counted ingredient form, a tag with a count
event.recipes.hbm.press('minecraft:emerald', {ingredient: 'minecraft:gold_block', count: 2}, 'flat')
// a material-and-shape input (steel ingots)
event.recipes.hbm.arc_welder('hbm:fuel_tank_large',
    [{material: 'steel', shape: 'ingot', count: 8}, {ingredient: '#c:plates/saturnite', count: 12}], 200, 5000)
// a pressurized fluid
event.recipes.hbm.chemical_plant('kjs.pressurized')
    .inputFluids([{type: 'hbm:steam', amount: 1000, pressure: 1}])
    .outputFluids([Fluid.of('minecraft:water', 1000)])
    .duration(100).power(100)
// an output with a chance
event.recipes.hbm.chemical_plant('kjs.lucky')
    .inputItems(['minecraft:gravel'])
    .outputItems([{id: 'minecraft:diamond', chance: 0.05}])
    .duration(20).power(10)

On KubeJS 7 and 8 a typed item output has no chance, so a chance always needs the plain object form.

You can also write a whole recipe as JSON with KubeJS’s own event.custom:

event.custom({
  type: 'hbm:press',
  stamp: 'flat',
  input: {ingredient: 'minecraft:jungle_log'},
  output: 'hbm:ball_resin'
}).id('kubejs:press_resin')

Golem Core also accepts a single plain object as the whole row (event.recipes.hbm.press({...})) for tables whose function has no one-argument form. That excludes the generic tables, whose one-argument form is (name); use event.custom there.

Removing and replacing

ServerEvents.recipes(event => {
  // by recipe id (HBM's ids are hbm:<table>/<name>)
  event.remove({id: 'hbm:press/ball_resin'})
  // by table and output
  event.remove({type: 'hbm:press', output: 'hbm:biomass_compressed'})
  // by input, in every HBM shredder recipe
  event.remove({type: 'hbm:shredder', input: 'minecraft:cobblestone'})
  // swap an input in every press recipe; the row keeps its count
  event.replaceInput({type: 'hbm:press'}, 'minecraft:jungle_log', 'minecraft:oak_log')
})

Fields that hold item or fluid inputs take part in input: filters and replaceInput; item and fluid outputs in output: filters and replaceOutput. A generic table’s icon takes part in neither.

When a replacement names one item and the original had a count (or a chance), the count and chance stay: replacing {ingredient: 'minecraft:jungle_log', count: 4} with 'minecraft:oak_log' gives 4 oak logs.

A row written in one of HBM’s special forms (a material-and-shape input, a weighted SILEX output, a pressurized fluid, an output with item data) is not matched by item or fluid filters. It is never rewritten either. Remove such a row by id.

A field that holds a fluid id alone (no amount, for example the refinery’s input) matches a fluid filter as one bucket.

A removal list in kubejs/data/<ns>/recipe_removals/ removes rows by id too, and also works without scripts. See Removing recipes.

Errors

A value the field cannot hold is a script error that names the table and the field, never a silently wrong recipe:

event.recipes.hbm.shredder('minecraft:gravel', '2x minecraft:cobblestone')
// hbm:shredder.input: this input takes no count (got 2)
event.recipes.hbm.cyclotron('hbm:powder_neodymium', 'hbm:powder_cerium', '#minecraft:logs', 25)
// hbm:cyclotron.particle: this input takes an item, not the tag #minecraft:logs

Other messages: empty ingredient: ..., empty item: ..., empty fluid: ... (an unknown id), takes one item or one tag, not ... (a multi-item ingredient), this output takes no chance: ..., this output takes no count.

Plain objects are not checked by the script function. A row that is wrong inside (a bad stamp name, an unknown material, a missing key in a plain object) fails when the table reads it, and the error shows up in the log like a broken datapack file’s.

Signatures

table(required...) shows the positional arguments. The optional fields follow in the order shown and are also builder methods (with camelCase forms for names with underscores). The value form of each field is in brackets where it is narrower than the defaults above. order (a number) is the row’s position in the machine’s list; rows without one come last. requires names the 1.7 config switches a row depends on; leave it out for a row that is always on.

Fluid tables

Table	Positional arguments	Optional
coker	output [stack], input [fluid id]	amount, solid_fuel_energy, byproduct [fluid stack], byproduct_type [fluid id]
compressor	output [fluid stack], input [fluid id], amount, duration	pressure
cracking	left [fluid stack], input [fluid id]	right [fluid stack]
radiolysis	left [fluid stack], input [fluid id]	right [fluid stack]
electrolyser_fluid	input [fluid id], amount	output1, output2 [fluid stack], duration, byproducts [list of stack]
pseudo_fluid	name, outputs [list of stack], consumed, produced	from [fluid id], next, high_speed
gas_centrifuge	outputs [list of stack], input [fluid stack], centrifuges	high_speed
hydrotreating	outputs [list of fluid stack], input [fluid id], hydrogen [fluid stack]
liquefaction	output [fluid stack], input [item or tag]
pyrolysis	duration	output_item [stack], output_fluid [fluid stack], input_item [ingredient], input_fluid [fluid stack], solid_fuel [fluid id], solid_fuel_energy
refinery	outputs [list of fluid stack], byproduct [stack], input [fluid id]
vacuum_distill	outputs [list of fluid stack], input [fluid id]
reforming	outputs [list of fluid stack], input [fluid id]
solidification	output [stack], input [fluid id]	amount, solid_fuel_energy
storage_drum	output [stack], input [item stack in], decay	divisor, liquid, gas
fraction	left [fluid stack], right [fluid stack], input [fluid id]

Item-processing tables

Table	Positional arguments	Optional
tool_conversion	tool, from (id), to (id)	materials [list of ingredient]
material_distribution	materials (raw JSON list), input [item or tag]
arc_furnace	input [ingredient]	solid_output [stack], molten_outputs (raw JSON list), order
crucible	name, outputs (raw JSON list), inputs (raw JSON list)	frequency, icon [stack], order
centrifuge	outputs [list of stack], input [ingredient]	requires, order
crystallizer	output [stack], input [ingredient], acid [fluid stack], duration	productivity, requires (one id), order
cyclotron	output [stack], input [ingredient], particle [item], antimatter	order
electrolyser_metal	molten (raw JSON list), input [item]	byproducts [list of stack], duration, order
rotary_furnace	output (raw JSON), duration, steam	inputs [list of ingredient], fluid [fluid stack], order
silex	outputs (raw JSON list), input [ingredient], laser_strength, fluid_produced, fluid_consumed	order
shredder	output [stack], input [item or tag]

Item tables

Table	Positional arguments	Optional
ammo_press	output [stack], grid [list of grid slot]	order
annihilator	milestones (raw JSON list)	item [item], tag [tag], fluid [fluid id], requires
arc_welder	output [stack], ingredients [list of ingredient], duration, consumption	fluid [fluid stack], order
breeder	output [stack], input [item], flux	order
combination	input [ingredient]	item_output [stack], fluid_output [fluid stack], order
lemegeton	output [item id], input [item or tag]	order
magic	result [item id], ingredients [list of item or tag]	order
particle_accelerator	output1 [stack], input1 [ingredient], input2 [ingredient], momentum	output2 [stack], order
pedestal	output [item id], grid [list of pedestal slot]	count, extra, set, order
press	output [stack], input [ingredient], stamp	order
sawmill	output [stack], input [ingredient]	crafting_result (raw JSON), sawdust_chance, hidden, order
soldering	output [stack], duration, consumption	toppings, pcb, solder [list of ingredient], fluid [fluid stack], requires, order
anvil_smithing	output [stack], left [ingredient], right [ingredient], tier	hot, shapeless, order
anvil_construction	outputs [list of output, always objects], inputs [list of ingredient], tier	tier_upper, overlay, requires, order

The press stamp is one of flat, plate, wire, circuit, c357, c44, c50, c9, printing1 to printing8. A pedestal slot is like a grid slot, but written "id" / "#tag", or {id, count} with a count.

Generic tables

These all share one schema: table(name), then any of these optional fields, in this order:

Field	Value
output_items / outputItems	list of output (item with count and chance)
input_items / inputItems	list of ingredient
output_fluids / outputFluids	list of fluid stack
input_fluids / inputFluids	list of fluid stack
duration	ticks (default 0)
power	HE per tick (default 0)
icon	stack, always written as an object
custom_localization / customLocalization	true: show the translation of name instead of the icon’s item name
name_wrapper / nameWrapper	translation key wrapped around the shown name
blueprint_pools / blueprintPools	list of pool names; a pooled recipe needs that blueprint, as in 1.7
auto_switch_group / autoSwitchGroup	group name
order	position in the machine’s list
input_items_expensive / inputItemsExpensive	list of ingredient used in expensive mode
input_fluids_expensive / inputFluidsExpensive	list of fluid stack used in expensive mode
blueprint_pools_528 / blueprintPools528	pools used in 528 mode
requires	list of config switches

Table	Extra optional fields
assembly_machine, blast_furnace, chemical_plant, difurnace, exposure_chamber, fluid_breeder, mixer, purex, precision_assembler, space_assembler, rock_mill, supercomputer	none
fusion	ignition_temp, output_temp, neutron_flux, red, green, blue
outgasser	fusion_only
plasma_forge	ignition_temp

name must be unique in the table. Machines save the selected recipe by it, and the recipe selector shows it. Always set duration and power: both default to 0.

event.recipes.hbm.chemical_plant('kjs.my_gas')
  .inputItems(['2x minecraft:iron_ingot'])
  .outputFluids([Fluid.of('hbm:oxygen', 500)])
  .duration(40)
  .power(100)
  .id('kubejs:chem_my_gas')

Tables from addon mods

An addon mod’s tables get schemas the same way, under its namespace (event.recipes.<namespace>.<table>). A table that states no fields still works: every JSON key becomes an optional raw-JSON field, so event.recipes.<namespace>.<table>({...}) or event.custom({...}) can write it.