Adaptive Mob Ai

A Minecraft 1.20.1 Forge mod that uses actual machine learning to make ALL vanilla mobs progressively smarter over time. Mobs learn from every interaction using a Deep Q-Network and evolve their behavior to create an ever-changing, adaptive world.

File Details

Adaptive-Mob-Ai-1.1.17-all.jar

  • B
  • Dec 15, 2025
  • 271.81 KB
  • 107
  • 1.20.1
  • Forge

File Name

Adaptive-Mob-Ai-1.1.17-all.jar

Supported Versions

  • 1.20.1

Curse Maven Snippet

Forge

implementation fg.deobf("curse.maven:adaptive-mob-ai-1402291:7335821")
Curse Maven does not yet support mods that have disabled 3rd party sharing

Learn more about Curse Maven

[1.1.17] - 2025-12-15

Added

  • AI Tier Progression System (Inspired by Hostile Neural Networks)

    • 5-tier intelligence system: UNTRAINED → LEARNING → TRAINED → EXPERT → MASTER
    • Mobs gain combat experience and level up their AI over time
    • Experience persists across mob deaths (global per-mob-type tracking)
    • Accuracy-based decision making (30% → 95% tactical success rate)
    • Dynamic think intervals (smarter mobs react faster)
    • Visual tier indicators with colored particles (green/blue/purple)
    • Tier-up celebration effects with enchantment glints
    • Federation sync for tier data (servers share AI progression globally)
    • Configuration options: enableTierProgression, visualTierIndicators, experienceRateMultiplier
  • Variant Family Grouping System

    • 12 mob families for cross-learning (zombie family, skeleton family, etc.)
    • Variant mobs share 50% experience with family members
    • Drowned learns from zombies, stray learns from skeletons, etc.
  • Mob-Specific Think Costs

    • Base intervals per mob type (zombies: 15t, skeletons: 20t, bosses: 30t)
    • Tier-based speed modifiers (MASTER mobs think 40% faster)
    • Performance optimized for complex mob AI
  • Cloudflare Worker Tier Endpoints

    • /api/tiers POST/GET for tier data upload/download
    • FederationCoordinator stores global tier progression in Durable Object
    • Merge strategy: keeps maximum experience per mob type

Changed

  • AI decision making now includes accuracy checks (lower tiers make tactical mistakes)
  • Think intervals now dynamic based on mob type and intelligence tier
  • Experience system tracks total combat outcomes per mob type globally

Technical

  • New AITier enum with 5 progression levels and tier metadata
  • TierVisualIndicators utility class for particle effects
  • Tier data sync methods in FederatedLearning and CloudflareAPIClient
  • Config loading enhanced to read tier progression settings
  • All systems maintain full backwards compatibility

[1.1.13] - 2025-12-14

Fixed

  • Critical: Mob freezing during federation uploads (blocking HTTP on server thread)
    • All uploads now use CompletableFuture.runAsync() fire-and-forget pattern
    • Dedicated HTTP executor ensures complete isolation from server thread
    • Data copying prevents concurrent modification during async upload

[1.1.8] - 2025-12-14

Added

  • Federated Learning v3.0.0: Complete rewrite with Durable Objects
    • Decisive round finality (10 min OR 10 contributors)
    • Forced startup pull ensures bootstrap traffic
    • GitHub flight recorder for debugging
    • Proof-of-life /status endpoint
    • Single source of truth coordinator

Changed

  • Client behavior: Now uses v3 API endpoints (/api/global, /api/upload)
  • Round closure: Hard limits prevent indefinite rounds
  • Startup flow: Guaranteed download on server start
  • Bootstrap uploads: First encounter triggers immediate upload

Fixed

  • No guaranteed bootstrap traffic (forced startup pull added)
  • Rounds never closing (time/contributor limits enforced)
  • No federation visibility (enhanced status endpoint)
  • Rate limit risk (logging once per round)

[1.1.7] - 2025-12-14

Added

  • GitHub logging system for federation debugging
  • GitHubLogger utility (async, non-blocking, safe)
  • Round finality rules in FederationCoordinator
  • Enhanced status endpoint with proof-of-life data

Fixed

  • Federation rounds staying open indefinitely
  • Lack of visibility into federation status

[1.1.6] - 2025-12-12

Added

  • Per-mob learning statistics in /amai stats command
  • Detailed breakdown of tactics by mob type

Changed

  • Enhanced stats display with mob-specific data

[1.1.5] - 2025-12-10

Added

  • Forced bootstrap uploads on first mob encounter
  • Initial upload system to establish federation connection
  • Heartbeat sync mechanism

Fixed

  • Servers not uploading data (bootstrap issue)
  • Federation connection not establishing reliably

[1.1.4] - 2025-12-09

Changed

  • Restored single JAR build (all dependencies bundled)
  • Simplified deployment (one file instead of two)

Fixed

  • Dependency management issues with split JARs

[1.1.3] - 2025-12-08

Added

  • Auto-increment version system for local builds
  • Build automation improvements

[1.1.2] - 2025-12-07

Added

  • Enhanced logging for federation operations
  • Better error messages

[1.1.0] - 2025-12-05

Added

  • Federated Learning System v2.0
    • Global model aggregation via Cloudflare Workers
    • Real-time tactic sharing across all servers
    • FedAvg algorithm for model updates
    • Automatic sync every 10 minutes

Changed

  • Moved from local-only learning to distributed system
  • Cloudflare Worker backend for coordination
  • Enhanced mob AI with global tactics

[1.0.0] - 2025-12-01

Added

  • Initial Release
  • Deep Q-Network for mob learning
  • Rule-based + learning hybrid AI
  • Support for 8 mob types:
    • Zombie, Skeleton, Creeper, Spider
    • Husk, Stray, Wither Skeleton, Enderman
  • MCA Reborn integration (optional)
    • Villager personality system
    • Mood-based dialogue
    • 40+ dialogue templates
  • Configuration system
    • Per-mob toggles
    • AI difficulty scaling (0.5 - 3.0)
    • Federation enable/disable
  • Commands:
    • /mcaai info - Mod status
    • /mcaai stats - AI statistics
    • /mcaai federation - Federation status
    • /mcaai test dialogue - Dialogue testing

Version Format

Format: [MAJOR.MINOR.PATCH]

  • MAJOR: Breaking changes, architecture rewrites
  • MINOR: New features, API changes
  • PATCH: Bug fixes, minor improvements