Universal Modding

Universal Modder: how to install it for every supported agent

Universal Modder is agent-agnostic: every supported coding agent gets the same Agent Skills set, the same fal MCP server for generated assets, and the same um CLI for recon, sprite work, backups, recording and packaging. Pick the section for your agent below, or read the six steps first if you are starting from zero.

Install steps

  1. Check the prerequisites. You need Git, Python 3.10+ and ffmpeg on PATH. uv is recommended for the um CLI, Blender is only needed for 3D-to-sprite rendering, and you should own the single-player game you want to mod.
  2. Pick your agent. Claude Code, Codex, Gemini CLI, VS Code / Copilot, Cursor and OpenCode all install the same skills, fal MCP server and um CLI. Claude Code and Codex have dedicated walkthroughs on this site.
  3. Run the install commands. Use the marketplace or extension command for your agent, or clone the repository and start the agent inside it. Every path ends with the same skills and tools available in the session.
  4. Configure asset generation (optional). Set FAL_KEY from your fal.ai dashboard so um fal and the fal MCP server can generate sprites, textures, audio and 3D. Without a key, um comfy can use a local ComfyUI server instead.
  5. Verify with the CLI. Run um --help in a new terminal and query the shared knowledge base with um kb search. Both work without opening a game, so problems surface before you risk anything.
  6. Make your first mod. Start the agent near a game you own and describe the mod in plain language. The mod-any-game skill searches field notes, picks a route and walks through the loop with checkpoints.

Per-agent install commands

Claude Code

/plugin marketplace add rehan-remade/universal-modder
/plugin install universal-modder@universal-modder

The plugin install brings the skills, fal MCP server and um CLI in one step. Full walkthrough with verification and troubleshooting: Universal Modder Claude Code.

Codex

codex plugin marketplace add rehan-remade/universal-modder
codex plugin add universal-modder@universal-modder

Codex reads the same skills and MCP configuration through its plugin system. Dedicated guide: Universal Modder Codex.

Gemini CLI

gemini extensions install https://github.com/rehan-remade/universal-modder

The extension ships the skills, the fal MCP server and the CLI together, so a single command gets a Gemini CLI session to the same starting line as the other agents.

VS Code / Copilot

Enable chat.plugins.enabled in your VS Code settings, run the command Chat: Install Plugin From Source, and enter the repository URL when prompted. Copilot then picks up the same skills from the plugin source.

Cursor

Install from the Cursor Marketplace, or clone the repository and open it in Cursor. Cursor reads AGENTS.md and .cursor/mcp.json from the working directory, so the clone path needs no extra configuration.

OpenCode

git clone https://github.com/rehan-remade/universal-modder
cd universal-modder
opencode

Cloning and running opencode inside the repository loads the skills and the fal MCP server through opencode.json. On OpenCode 2 you can instead run opencode plugin add with the repository path shown upstream — note that the plugin install does not put um on PATH, so install the CLI separately.

Skills only (any agent)

npx skills add https://github.com/rehan-remade/universal-modder

This installs just the Agent Skills format files for agents that do not have a plugin system, without the MCP server or the CLI.

Anything else

git clone https://github.com/rehan-remade/universal-modder

Cloning gives every agent the same foundation: skills in .agents/skills and .claude/skills, instructions in AGENTS.md, and MCP configuration per agent. Start your agent inside the clone and it finds them on its own.

What actually gets installed

However you install, the package contains twelve skills. mod-any-game is the entry point that runs the whole loop and carries the engine playbooks. game-recon fingerprints your install and produces a modding plan. game-research-websearch digs through archives and APIs when forums die. reverse-engineering drives ILSpy, Ghidra, Cheat Engine, Frida and RenderDoc. fal-assets and asset-pipeline generate and fit art, audio and 3D to engine-exact formats. game-automation and showcase-video handle in-game testing and recording. mashup-mods covers game-inside-a-game projects, port-a-mod moves existing mods across loaders, publish-mod packages for release, and share-field-notes writes the notes the next agent will read.

Requirements and safety, in one place

The toolkit works on games you own: single-player titles, multiplayer you host yourself, and offline modes. It refuses to inject into online games with anti-cheat, refuses cheats against other players, and never ships game files or decompiled code. Save folders are backed up before edits, processes are killed by PID only, and the agent asks before driving your input, installing loaders into game folders, or publishing a mod.

As for hardware: a normal PC or a WSL environment is enough. GPU-side work like ComfyUI is optional and only matters if you want local image generation instead of fal.

Verify before you mod

um --help
um kb search "terraria"

If both commands answer, the install is complete and the agent can search prior field notes before it writes a single line of mod code. Head to the first mod walkthrough when you are ready to try it on a game.