Skip to content

Rigs Docs

Generate images by chatting with your AI agent

Rigs turns your local ComfyUI into a conversational creative tool. Instead of wrestling with node editors, you describe what you want in natural language — through Discord, Telegram, iMessage, Slack, or any channel your OpenClaw agent lives on — and Rigs runs your workflows and delivers results right in the chat.

You: "a samurai standing in cherry blossom rain, cinematic lighting"
Agent: 🖼️ Running FLUX Klein T2I...
Agent: [image attached]

Every generation lands in your local dashboard gallery automatically. From there, you can publish workflows to the Rigs marketplace, where anyone can pay to run them on your GPU — turning your creative setup into a revenue stream.

The Three Layers

1. Chat-Powered Generation (the magic)

Talk to your OpenClaw agent from anywhere. It picks the right workflow, fills in parameters, queues the job on your local ComfyUI, and sends back the result. No browser needed.

  • Works on Discord, Telegram, iMessage, Slack, Signal, WhatsApp, IRC — anywhere OpenClaw runs
  • AI-Enhanced Workflows interpret simple prompts through your custom style guide
  • Supports text-to-image, image-to-image, video, background removal — any ComfyUI workflow

2. Local Dashboard & CLI

A web dashboard at localhost:8787 and a rigs CLI for full control:

  • Servers — manage multiple ComfyUI instances
  • Workflows — import JSON, classify inputs/outputs, set cover art, configure AI enhancement
  • Pipelines — chain workflows into multi-step runs
  • Gallery — every output stored locally with metadata, filtering, downloads
  • Network — publish workflows, set pricing, connect to the marketplace

3. Rigs (rigs.run)

A decentralized GPU marketplace where your workflows earn money:

  • Consumers browse and run workflows at rigs.run
  • Jobs execute on your GPU — no uploads, no cloud, your IP stays local
  • Wallet system with Stripe payments and provider payouts
  • Live progress streaming and preview images over WebSocket
  • Metadata protection strips workflow JSON from outputs by default

How It All Connects

Discord / Telegram / iMessage / Slack
        │
        ▼
   OpenClaw Agent (your AI assistant)
        │
        ▼
   Rigs Skill (dashboard + CLI)
        │
        ▼
   ComfyUI Server (your GPU)
        │
        ▼
   Results → Chat + Gallery + rigs.run
  1. You message your agent: "cyberpunk cat, neon rain"
  2. OpenClaw routes to the Rigs skill
  3. Rigs picks the right workflow, queues the job on ComfyUI
  4. The output lands in your chat and your local gallery
  5. If the workflow is published, consumers on rigs.run can run it too — and you earn per run
What Where
Install OpenClaw + Rigs Getting Started · Installation
Run your first workflow Getting Started · First Workflow
Dashboard walkthrough Getting Started · Dashboard
Set up AI-Enhanced Workflows Providers · AI-Enhanced Workflows
Become a provider Providers · Setup
API reference API · Overview

Source

Repo What
mitchellfox/rigs Local skill (dashboard + CLI)