About OpenShopFloor
A personal, non-commercial open source project.
What is OpenShopFloor?
OpenShopFloor is an open source sandbox for manufacturing AI. It provides a live factory simulation with 118 MCP tools, a visual Node-RED flow editor, a TypeScript agent SDK, and pre-built multi-agent analysis systems.
The goal is simple: give anyone a real factory environment to explore, experiment, and build AI agents for manufacturing — without needing access to a physical production facility.
Why does this exist?
Manufacturing generates enormous amounts of data — OEE metrics, quality measurements, material flows, energy consumption, maintenance logs. AI has the potential to transform how factories operate. But there is a problem:
You can't experiment with factory AI if you don't have a factory.
OpenShopFloor solves this by providing a realistic, always-running manufacturing simulation with the same data structures, complexity, and edge cases you would find in a real production environment. CNC machines, injection molding, assembly lines, quality control, tool management, ERP — all accessible through standardized MCP (Model Context Protocol) tools.
Open Source — AGPL-3.0
OpenShopFloor is a personal project, not a commercial product. There is no company behind it, no venture funding, no paid tier, no hidden monetization.
The platform is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). This means you are free to use, modify, and distribute the code — but if you run a modified version as a service, you must publish your changes under the same license.
What is open source (AGPL-3.0):
- Frontend (Next.js web application)
- Gateway (API server, Node-RED integration)
- Flow engine (BFS execution, multi-input/output)
- Agent sandbox (isolated-vm TypeScript runtime)
- All custom Node-RED nodes
What is NOT open source:
- Factory simulation (proprietary, hosted as a free service)
- MCP server data layer (proprietary)
The factory simulation is provided free of charge for experimentation. It took significant domain expertise to build and is not part of the open source release. You can connect your own data sources via MCP instead.
I built this because I believe manufacturing AI should be accessible to everyone — not just companies that can afford six-figure platform licenses.
The Simulated Factory
CNC Machines
Milling and turning centers with realistic cycle times, tool wear, and OEE data.
Injection Molding
SGM machines with 97 process parameters, cavity balance, and SPC monitoring.
Assembly Lines
Multi-station assembly with pre-assembly cells, buffer management, and test field.
Full ERP System
Work orders, BOM, stock, purchases, customer orders, and MRP planning.
Quality Management
Defect tracking, SPC, audits, corrective actions, and customer complaints.
Tool & Warehouse Management
Tool inventory, calibration, maintenance schedules, stock movements.
What You Can Do
- 01Explore — Browse the live factory dashboard, query data through the chat, inspect all 118 MCP tools.
- 02Run built-in agents — Try multi-agent deep analysis systems for OEE, quality, delivery, and revenue optimization.
- 03Build visual flows — Design AI workflows in the Node-RED editor with custom nodes for MCP, LLM, context aggregation, and more.
- 04Write code agents — Deploy TypeScript agents from GitHub with full SDK access to MCP tools, LLM, and storage.
- 05Experiment freely — Break things, try wild ideas, test hypotheses. It's a sandbox — that's what it's for.