Edit photos by describing the change, on your own GPU. A free, open-source alternative to Nano Banana and ChatGPT image editing: photos stay on your computer, no daily limit. One installer for Windows, Mac and Linux sets up the best open model (Qwen-Image 2.1). MCP server for AI agents.
TypeScript ai-photo-editorcomfyui-alternativegenerative-fillimage-editingimage-generationinpaintinglinuxlocal-aimacosmcpmcp-servernano-banana-alternativeofflinephoto-editorprivacyqwen-imageqwen-image-2-1stable-diffusion-cppvulkanwindows
// Text printed by `kokon-mcp --help`. Kept apart from index.ts so it can be tested without
// starting the server. The environment variables listed here are read in paths.ts,
// service.ts and index.ts; update this text when one is added or renamed.
export function helpText(version: string): string {
return `kokon-mcp ${version}
MCP server for Kokon. It speaks the Model Context Protocol over stdio, so an MCP client
(Claude Code, Cursor, Claude Desktop...) starts it; running it by hand just waits for input.
It needs the Kokon desktop app installed (https://kokon.app) and starts the local engine on demand.
Usage:
kokon-mcp start the MCP server on stdio
kokon-mcp --version print the version and exit
kokon-mcp --help print this text and exit
Register it with Claude Code:
claude mcp add --scope user kokon -- npx -y kokon-mcp
Environment variables:
QLS_MCP_ALLOWED_DIRS extra folders the tools may read and write, separated by semicolons
QLS_SERVICE_EXE path to kokon-service, if Kokon is installed in an unusual place
QLS_DATA_DIR Kokon's data folder (default: the app's KokonData folder)
CLAUDE_PROJECT_DIR project folder the tools may use (set by Claude Code; default: the
working directory)
`;
}