Braincraft Mod Notes

What it does

• Runs a server-side artificial life sandbox with hundreds of lightweight agents (no ML). Each agent owns a modular brain built from neuron clusters and neuromodulators (dopamine, cortisol, oxytocin, acetylcholine) that learn locally and adapt.
• Agents seek food/water gradients, flee hazards, align and cohere into herds, share improvised words, and reproduce when resources and stress allow. Behavior emerges from survival signals, social pressure, and language reinforcement rather than scripted routines.
• Virtual resource/hazard fields drift over time, forcing migration and new social clusters. Newborns inherit lexicon fragments and state from parents, keeping cultures alive.

How to use in-game

• Load on a Fabric server for Minecraft 1.20.1. The simulation starts automatically around the world spawn.
• Commands (all users):
  • /braincraft summary — live population and mood snapshot plus the current dominant word.
  • /braincraft agent <id> — inspect an individual agent’s energy, stress, attunement, novelty, fertility, position, velocity, and top vocabulary.
  • /braincraft spawn — seed a small cluster of new agents if headroom remains.

Implementation details

• Entry point: dev.braincraft.BraincraftMod registers the simulation, server tick loop, and commands.
• Core simulation lives in dev.braincraft.sim.BraincraftSimulation with agents, spatial hashing for neighbor sense, drifting resource layers, and message delivery for language.
• Brains combine neuron clusters (survival, social, exploration, language, reproduction) with modulators; weights update locally each tick based on reward/stress signals for non-scripted adaptation.

Known limitations / future ideas

• Agents are virtual and do not spawn visible entities; feedback is via commands only.
• Interaction with the physical world is limited to the spawn-centered coordinate space; tying stimuli to actual blocks/biomes or player proximity could deepen emergence.
• Language is lightweight (syllable strings); future work could add simple grammar tokens or environmental tagging.