The WRG MCP Server acts as a specialized digital architect for first-person shooter (FPS) game development, handling the complex physics of how weapons "kick" when fired. Instead of requiring developers to manually program every subtle movement of a gun's barrel, this tool generates realistic recoil patterns instantly. It simplifies the creative process by allowing anyone to define how a machine gun, pistol, or shotgun should behave with just a few simple commands, making it much easier to fine-tune the feel of a game. Technically, the server operates by exposing specific HTTP endpoints for Weapon Recoil Generation (WRG) and Recoil Visualization (RVZ). It includes dedicated functions such as `machinegun_recoil_points` and `pistol_recoil_points` that calculate precise coordinate trajectories based on the number of shots fired. To make this data actionable, the built-in visualizer can transform these raw coordinates into 2D scatter plots, providing an immediate graphical representation of a weapon's spray pattern for analysis and debugging. For developers integrating with AI systems like Claude, this MCP server turns an LLM into a powerful game-balancing assistant. By connecting these tools, an AI can not only suggest gameplay improvements but also generate the actual mathematical data and visual proofs to back them up. This integration creates a seamless workflow where developers can describe a desired gameplay experience in plain English and receive fully realized, testable recoil data in return, significantly accelerating the prototyping and iteration phases of game design.
Category: Design, Media & Creative
Tags: fps, game-balancing, gamedev, physics, recoil
bash git clone https://github.com/Hyeongseob91/mcp-server.git cd mcp-server pip install -r requirements.txt To run the server locally for development: bash uvicorn main:app --reload --host 0.0.0.0 --port 8000~/Library/Application Support/Claude/claude_desktop_config.json - Windows: %APPDATA%/Claude/claude_desktop_config.json JSON Configuration: json { "mcpServers": { "wrg": { "command": "python /path/to/mcp-server/main.py --http" } } } Note: Replace /path/to/mcp-server/main.py with the actual absolute path to the main.py file on your system.machinegun_recoil_points(shots: int): Generates recoil trajectory data for a machine gun based on the specified number of shots. - pistol_recoil_points(shots: int): Generates recoil trajectory data for a pistol based on the specified number of shots. - shotgun_recoil_points(shots: int): Generates recoil trajectory data for a shotgun based on the specified number of shots. - plot_recoil_pattern(data: Tuple[List[float], List[float]]): Visualizes input recoil coordinate data as a 2D scatter plot.Y coordinate offsets for weapon spray patterns or building custom internal tools just to see how a new weapon feels. This slows down the creative process when trying to decide if a new "Heavy Machine Gun" should have a vertical or zig-zag recoil. Solution: Using the WRG MCP, a designer can use Claude to instantly generate and visualize different recoil trajectories. They can iterate on the "feel" of a weapon in a conversational interface before writing a single line of game engine code. Example: A designer asks Claude: "Generate a 30-shot machine gun recoil pattern and show me the plot. Does this look too difficult for a beginner player?" Claude calls machinegun_recoil_points(30) and plot_recoil_pattern(), displaying a scatter plot that the designer can immediately evaluate.
Part of MCP Servers
WRG MCP Server is a Model Context Protocol service that exposes HTTP endpoints for weapon recoil generation and 2D scatter plot visualization. It lets developers calculate weapon kick patterns for pistols, shotguns, and machine guns directly through AI agents.
Clone the repository from GitHub, navigate to the folder, and run pip install -r requirements.txt. You can then run the service locally using Uvicorn or configure your MCP client to start the script with the http flag.
The server includes machinegun_recoil_points, pistol_recoil_points, and shotgun_recoil_points for generating recoil coordinate datasets based on shot count, as well as plot_recoil_pattern for rendering coordinate lists as 2D scatter plots.
The server works with Claude Desktop on macOS and Windows by adding the command execution string to your claude_desktop_config.json file. It also supports remote deployments to platforms like Smithery via FastMCP.