Console Combat ABot Logic

I already have the capture pipeline in place: AI-driven computer-vision reads the console’s HDMI feed, pushes decisions through Titan Two and a VIGEm virtual controller, then injects the inputs back to the console. What I still need is the real “brain” of the system—a complete ABot logic layer focused purely on combat assistance.

Scope of work
• Combat Assistance Only: no resource gathering or NPC chat—just fight-centric behaviour.
• Features required: reliable auto-targeting plus streamlined skill automation (cool-downs, combos, burst rotations, etc.).
• Multi-Game Flexibility: the logic must be structured so I can drop in profiles for a variety of titles without rewriting the core. Think game-agnostic state machine or modular behaviour tree rather than hard-coded offsets.

Tech context
• Runs headless on Windows.
• Feeds come from an ML vision module (OpenCV + YOLO). Your layer simply calls my existing detection API and returns controller inputs.
• Inputs are issued through Titan Two scripting or direct VIGEm calls—whichever is cleaner for you.

Deliverables
1. Well-commented source (Python, C++, W/GPC script) implementing the combat logic.
2. At least one working example profile that demonstrates auto-targeting and skill rotation in a live console session.
3. Clear instructions for adding new game profiles and adjusting thresholds.
4. Short video or live demo proving the example profile functions in real-time.

Acceptance criteria
•