comfyui-mcp-server and comfy-mcp-server
These are complementary tools serving different deployment architectures: one wraps a local ComfyUI instance as an MCP server, while the other uses FastMCP to connect to a remote ComfyUI backend, allowing users to choose between self-contained local setups versus distributed client-server configurations.
About comfyui-mcp-server
joenorton/comfyui-mcp-server
lightweight Python-based MCP (Model Context Protocol) server for local ComfyUI
Exposes ComfyUI's generation capabilities (images, audio, video) through MCP's streamable HTTP transport with stateful job management—enabling AI agents to poll async operations, iterate via `regenerate`, and inspect outputs directly. Auto-discovers custom workflows from JSON files with parameterized placeholders, while optional asset publishing with deterministic compression integrates generated media into web projects with manifest tracking.
About comfy-mcp-server
lalanikarim/comfy-mcp-server
A server using FastMCP framework to generate images based on prompts via a remote Comfy server.
Exposes ComfyUI workflows as MCP tools with configurable prompt and output nodes, enabling Claude and other MCP clients to trigger image generation with polling-based job tracking. Optionally integrates with Ollama for autonomous prompt generation from topics. Designed for stdio transport integration with Claude Desktop, abstracting workflow complexity through environment-based configuration and template substitution.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work