Skip to content

Track A outline — MP Bot AI (UT2004-style)

Status: Outline stub only · 2026-10-26
Depends on: Track D substrate
Not started: Full design spec · Unreal graphs · mode strategy tables

Multiplayer arena / match bots that feel like UT2004 — difficulty ladder, aggression, accuracy, reaction, team play per game mode. Separate product brain from monsters and campaign patrol.

  • GFP TFAAI, ATFAAIController, fuzzy-in-LD, Squad LD + Agent LD
  • DA_TFA_AIDifficultyProfile rows with UT2004 semantics (labels and feel locked in full Track A spec)
  • Bot Squad / Agent LD graph family
  • Intent vocabulary for Ladder / Progressive MP / fill (objectives, flag, hold, roam, hunt)
  • Per-mode team strategy tables
  • UT2004 difficulty ladder (exact mapping — full spec)
  • Competitive Ladder fill / practice
  • Progressive Multiplayer when bots are authored
  • Not Demo sandbox C (that is Track B)
  1. Lock UT2004 difficulty table → DA_TFA_AIDifficultyProfile columns
  2. Enumerate Year-1 modes that need bot teams
  3. Author Squad intents per mode
  4. Acceptance: human can tell Novice vs Godlike without a HUD cheat