Implements a BeamNG.drive mod that integrates with the existing MSAgent-AI Named Pipe API to provide real-time AI commentary on driving events (vehicle changes, crashes, damage, surroundings).

Architecture

BeamNG.drive (Lua) → HTTP → Bridge Server (Python) → Named Pipe → MSAgent-AI (C#/.NET)

• BeamNG Mod (beamng-mod/): Lua extension that monitors vehicle state, detects crashes via sudden deceleration, tracks damage accumulation, and sends events via HTTP
• Bridge Server (beamng-bridge/bridge.py): Flask HTTP server that translates game events into Named Pipe commands (CHAT: prompts) for the existing MSAgent-AI pipeline
• Integration: Leverages merged MSAgent-AI desktop app's PipelineServer.cs and Ollama AI backend

Event Detection

Crash detection uses configurable thresholds:

local crashSpeedDelta = 30  -- km/h speed loss threshold
local crashEndSpeed = 10    -- max final speed after crash

if speedDelta > crashSpeedDelta and env.speed < crashEndSpeed then
  sendToAI("/crash", {vehicle_name = name, speed_before = prevSpeed})
end

Damage differentiation:

• Major damage (>0.1): /dent endpoint
• Minor damage (<0.1): /scratch endpoint

Setup

Windows users run beamng-bridge/setup.bat and start.bat, then install mod to %LOCALAPPDATA%\BeamNG.drive\[version]\mods\msagent_ai\.

All communication is localhost-only (HTTP to 127.0.0.1:5000, Named Pipe IPC).

Documentation

• QUICKSTART.md: Installation steps
• ARCHITECTURE.md: Data flow diagrams, extension points
• beamng-mod/README.md: Mod configuration, troubleshooting

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.