MCP Server
Generate game assets from a conversation with your AI assistant.
Included with the Pro and Studio plans. See what a generation costs
Using a coding agent? Give it the Ludo CLI instead. It's the recommended way to connect agents to Ludo. It uses far fewer tokens, because the agent loads only the commands it needs, not the schema of every tool. And it handles files for you: local files go in by path and results are saved to disk, so the agent never has to push base64 through its context.
Set up in 3 steps
Subscribe to Pro or Studio
API and MCP access is included on the Pro and Studio plans.
Get Your API Key
Log in at app.ludo.ai, open API Keys from the sidebar or the user menu, name your key and click Create key. Copy it right away: it is shown only once.
Connect your client
Add the Ludo MCP server to your assistant with one of the setups below.
MCP (Model Context Protocol)
Connect Ludo.ai to AI assistants like Claude using the Model Context Protocol. Generate game assets, animate sprites, and create visuals directly through natural language conversations.
Run this command to add Ludo.ai as an MCP server:
Add Ludo.ai to OpenAI Codex by editing ~/.codex/config.toml:
[mcp_servers.ludo-ai]
url = "https://mcp.ludo.ai/mcp"
[mcp_servers.ludo-ai.http_headers]
Authorization = "ApiKey YOUR_API_KEY"Replace YOUR_API_KEY with your Ludo.ai API key, then restart Codex so it reconnects with the new configuration.
Add Ludo.ai to Cursor using either the Settings UI or manual configuration:
Option 1: Cursor Settings UI
- Open Cursor Settings (
Cmd+,on Mac orCtrl+,on Windows/Linux) - Go to Tools & MCP section and click New MCP Server
- Set Server Name to
ludo-ai - Set Transport Type to
http - Set URL to
https://mcp.ludo.ai/mcp - Add a header:
Authorizationwith valueApiKey YOUR_API_KEY - Save and restart Cursor
Option 2: Manual Configuration
Edit your Cursor MCP config file at ~/.cursor/mcp.json (or .cursor/mcp.json in your project):
{
"mcpServers": {
"ludo-ai": {
"url": "https://mcp.ludo.ai/mcp",
"transport": "http",
"headers": {
"Authorization": "ApiKey YOUR_API_KEY"
}
}
}
}Replace YOUR_API_KEY with your Ludo.ai API key, then restart Cursor.
Option 3: mcp-remote proxy (fallback)
If the direct HTTP connection above doesn't connect, use the mcp-remote proxy instead. It runs locally via npx and bridges to our server. Add this to ~/.cursor/mcp.json:
{
"mcpServers": {
"ludo-ai": {
"command": "npx",
"args": [
"-y",
"mcp-remote@latest",
"https://mcp.ludo.ai/mcp",
"--transport",
"http-only",
"--header",
"Authorization: ApiKey YOUR_API_KEY"
]
}
}
}Replace YOUR_API_KEY with your Ludo.ai API key, then restart Cursor. Requires Node.js installed (for npx).
CLI or MCP?
For AI agents, use the CLI: it uses far fewer tokens, and files go in by path instead of base64. MCP is for chat assistants that can't run commands. Explore the CLI
| Ludo CLI | MCP server | |
|---|---|---|
| Setup | npm install, then ludo login in the browser | Add the server and API key to each client |
| Files | Pass local paths; results saved to disk | URLs and base64; results expire after 7 days unless downloaded |
| Token use | Low: reads only the commands it uses, and local files go in by path | High: loads every tool schema (30+) up front, and a local file has to be written out as base64 inside the tool call |
| Works in | Any terminal, any agent that can run commands, scripts and CI | MCP-enabled clients |
| Best for | Coding agents (recommended), building a game in a repo, batch work, automation | Chat assistants that can't run commands |
Build with Ludo from your assistant
Ask for sprites, animations, 3D and audio in plain words.