🏭 TRUMPF Oseon MCP Server

Connect Claude AI to your TRUMPF Oseon manufacturing system. Ask questions in natural language, get real-time production insights.

Quick Start

Requirements: Python 3.10+, Claude Desktop, TRUMPF Oseon API access

# Install uv package manager
curl -LsSf https://astral.sh/uv/install.sh | sh  # macOS/Linux
# or: irm https://astral.sh/uv/install.ps1 | iex  # Windows

# Setup
git clone https://github.com/SheetMetalConnect/Oseon-MCP.git
cd Oseon-MCP
uv sync

# Configure
cp env.example .env
# Edit .env with your Oseon server details

# Install to Claude Desktop
uv run mcp install -m trumpf_oseon_mcp --name "TRUMPF Oseon"

Test: Open Claude Desktop and ask: "Show me today's production status"

What You Can Ask

Orders & Production:

• "Show me order 400139 with its production jobs"
• "What orders are overdue?"
• "Find all orders for customer ACME"

Analysis:

• "What's in production right now?"
• "Show me completed orders this week"
• "Which orders are released but not started?"

Dashboards:

• "Give me a production summary for the last 7 days"

Features

• ✅ Read-Only - Safe, no data modifications
• 📄 Pagination - Efficient handling of large datasets
• 🔍 Quality Filtering - Excludes template/test orders
• 📅 Smart Defaults - 12-month rolling window
• 🎯 15 Tools - Customer orders, production orders, dashboards

Architecture

src/trumpf_oseon_mcp/
├── api/          # HTTP client
├── models/       # Data types
├── utils/        # Filters, formatters, pagination
└── tools/        # MCP tools (customer_orders, production_orders, dashboards)

Key Components:

• OseonAPIClient: Centralized HTTP client for Oseon API
• Customer Orders: 6 read-only tools with pagination
• Production Orders: 7 read-only tools with status filtering
• Dashboards: 2 demo tools for quick analysis

See ARCHITECTURE.md for detailed design.

Examples

# Get customer orders (auto-paginated, last 12 months)
get_customer_orders()

# Search orders
search_customer_orders(search_term="ACME%")

# Filter by status
get_customer_orders_by_status(status="COMPLETED")

# Get production orders in progress
get_in_progress_production_orders()

# Quick dashboard
get_production_summary(days_back=7)

Development

See CONTRIBUTING.md for development setup.

# Install with dev dependencies
uv sync --dev

# Run tests
uv run python tests/test_unified_system.py

How It Works

[TRUMPF Oseon] --> [MCP Server (local)] --> [Claude Desktop]
                         ↓
                  Your Credentials
                  Read-Only Access

Privacy: MCP server runs locally. Only chat messages go to Claude; your manufacturing data stays on your network.

About

Built by Luke van Enkhuizen to demonstrate AI + manufacturing integration.

No TRUMPF affiliation - Educational project using public API documentation.

MIT License - Use freely | v2.0.0 - Modular architecture