Arthur Longbottom (artokun)
Arthur Longbottom (artokun) is an individual software engineer and solo developer/creator, not a separately documented company. His principal current project, ComfyUI MCP, is a local-first, agent-native control plane for ComfyUI: an MCP server and sidebar agent that lets an AI assistant generate media, author and run workflows, manage models and custom nodes, and edit a live ComfyUI graph in natural language.
At a Glance
- ComfyUI artists and creative technologists
- Developers building AI-media workflows and MCP clients
- Users with local GPUs who want an agent to operate ComfyUI
- Remote-GPU and RunPod users
- +2 more
AI Tools by Arthur Longbottom (artokun)
(1)
ComfyUI MCP
MCP Server for ComfyUI
Discussions
No discussions yet
Be the first to start a discussion about Arthur Longbottom (artokun)
Latest News
comfy-kitchen in ComfyUI: GPU capability reporting and reversible fp8/INT8/NVFP4 optimization
Security patch, reconnect self-healing, OpenRouter full catalog and third-party host protocol
RunPod connector and cloud LoRA training released
GPT-5.6 provider support, train_* LoRA tools and mechanically enforced Blind mode
Products & Services
Open-source MIT MCP server and Claude Code plugin for local, remote or Comfy Cloud ComfyUI. It provides tools for image, video and audio generation, workflow authoring/execution, asset handling, model search/download/removal, custom-node management, installation/environment management, process control, RunPod and LoRA-training workflows.
A ComfyUI sidebar agent that drives the live graph on Claude, ChatGPT, Gemini, local Ollama/LM Studio/llama.cpp or hosted OpenAI-compatible models. It supports natural-language graph edits, workflow and pack loading, spatial layout, rollback, multi-tab sessions, activity cards and cost confirmation for paid API nodes.
Flutter iOS/Android companion that pairs with the Agent Panel to chat with the agent, monitor renders and queues, browse CivitAI, choose providers, control RunPod hosts and mirror/drive a desktop tab.
GPU-performance tooling that reports what a machine can run, identifies graph operations still using default dtype and can apply or revert fp8, INT8 attention and NVFP4 optimizations.
Market Position
ComfyUI MCP positions itself against thin prompt-to-image MCP connectors by offering a full control plane that edits graphs, manages models and nodes, runs and debugs workflows, and supports many LLM backends. It is differentiated from Comfy-Org's hosted Comfy Cloud MCP and in-app agent by being local-first/self-hostable and able to target a user's own Mac, Linux or Windows installation, while still supporting remote and community cloud modes. Its trade-off is that users remain responsible for their ComfyUI environment and GPU setup.
Leadership
Founders
Arthur R. Longbottom (artokun)
Software engineer and creative technologist. His public profiles describe experience with 3D, AI, ComfyUI, TypeScript and WebGL; Wellfound identifies him as founder of Digital Pine and lists prior work/experience associated with Apple, TikTok, Adidas, Prada, Oracle and LG. He is also a regular contributor to Comfy-Org's ComfyUI frontend and Python core repositories.
Executive Team
Arthur R. Longbottom (artokun)
Creator and maintainer
Software engineer and creative technologist; public profiles cite 3D, AI, TypeScript and WebGL expertise, Digital Pine founding, and contributions to Comfy-Org's frontend and Python core.
Founding Story
The documented motivation for ComfyUI MCP is to provide more than a thin prompt-to-image connector: a full agent that can inspect and operate a user's ComfyUI installation, author and edit graphs node by node, execute and iterate workflows, manage models and custom nodes, and work locally, remotely or with Comfy Cloud. The project is explicitly presented as built and maintained by Arthur Longbottom under the artokun identity.
Business Model
Revenue Model
The documented project is open-source MIT software distributed through GitHub and npm. The software can use users' existing Claude, ChatGPT, Gemini or other subscriptions, local models, or hosted API providers; the public materials do not describe a paid ComfyUI MCP subscription or usage-based price for the project itself.
Target Markets
- ComfyUI artists and creative technologists
- Developers building AI-media workflows and MCP clients
- Users with local GPUs who want an agent to operate ComfyUI
- Remote-GPU and RunPod users
- Teams or individuals wanting provider-neutral, local-model or offline agent operation
- Mobile users who need remote visibility and control of ComfyUI
- AI-assisted image generation and editing
- Video and audio generation and post-production
- Natural-language construction, repair and iteration of ComfyUI workflows
- Managing local ComfyUI models, custom nodes and runtime resources
- Running ComfyUI on a local workstation, LAN/VPS, RunPod or cloud target
- Using local/free LLMs offline through Ollama or other OpenAI-compatible runtimes