abidoo22/pixelorama-mcp

Pixelorama-MCP, also called pix-MCP, is a small server that lets an AI assistant draw pixel art inside Pixelorama, a free pixel art editor. You type a request in plain English, like asking for a 32 by 32 sword with a golden blade and brown handle, and the AI uses this server to do the actual drawing inside the editor. The idea is that you do not need any drawing skills yourself. The project sits in the middle of two programs. On one side is the AI client, such as Claude Desktop, Cursor, or another tool that speaks the Model Context Protocol, a shared way for AI assistants to talk to outside software. On the other side is Pixelorama, with a small plugin installed inside it. The pix-MCP server is written in TypeScript and runs on Node.js. It receives instructions from the AI and forwards them to the Pixelorama plugin over a local HTTP connection. The README lists more than thirty drawing tools that the AI can call, covering pixels, lines, rectangles, ellipses, polygons, and flood fill. It can also create and rename layers, change opacity and blend modes, set colours, build palettes, and pick colours from the canvas. There is animation support for adding frames, setting durations, and exporting GIFs, plus selection and transform actions like move, resize, flip, and rotate. A helper called generate_sprite chains several of these steps together from a single text prompt. To use it, you clone the repository, install Node.js packages, build the server, then install a plugin file into Pixelorama and restart the editor. You then point your AI client at the built server file through a configuration entry. The README links to a getting started guide, a tool reference, a plugin setup guide, and an agentic drawing playbook for developers who want to write their own drawing scripts. The licence is MIT.