PTC has launched the FeatureScript MCP Server in Onshape, giving engineers a way to create custom CAD features using natural language and AI. The capability, available through the Onshape Labs early-access program, lets engineers describe what they need in plain language, and AI helps build, test, debug, and refine the corresponding FeatureScript code.
FeatureScript is Onshape's programming language for CAD automation, used to capture engineering knowledge, automate repetitive design tasks, and build custom design tools. The new server extends that capability to large language models including Claude, ChatGPT, and Gemini through the Model Context Protocol, or MCP. Rather than generating one-off CAD models, engineers can use natural-language instructions to create reusable FeatureScript capabilities that can be shared across teams and projects.
The approach is designed to make CAD customization more accessible to engineers who don't routinely work with FeatureScript code. It also supports PTC's Intelligent Product Lifecycle strategy, which focuses on establishing a consistent product data foundation across engineering and other enterprise functions.
From announced to available
PTC first outlined the FeatureScript MCP Server when it launched Onshape Labs in July 2026, alongside AI agents, drawing validation, AI-assisted rendering, and robotics simulation. This release moves it from an announced capability into availability within the program.
The server builds on PTC's broader AI expansion in Onshape. Its latest Onshape AI Advisor release introduced real-time design guidance and outlined agent-based workflows, including FeatureScript generation and automated engineering operations. PTC has taken a similar direction in its other products: Creo 13 and Creo+ 13.3, released earlier this year, included Creo AI Assistant for embedded guidance within CAD workflows.
Connecting LLMs to CAD code
Rather than limiting AI use to generating individual CAD models, the FeatureScript MCP Server connects LLMs directly with FeatureScript code. Engineers can describe the functionality they need, and AI can help build, test, debug, and refine the corresponding custom feature.
Once developed, these custom features can be shared across teams and projects, letting organizations turn established engineering practices into repeatable design tools. PTC's approach extends AI from design assistance toward the creation of reusable engineering automation.
Why this matters for product development professionals
For engineers and product development teams, the practical value is in capturing engineering knowledge as reusable tools rather than one-off outputs. AI-generated designs still require engineering review to ensure they meet functional, manufacturing, safety, and regulatory requirements - but the ability to create custom CAD features without deep FeatureScript expertise lowers the barrier to automation. Teams that combine AI-assisted code generation with their existing product context can build capabilities that persist beyond a single design session, which is where the real efficiency gains come from.
PTC's trajectory across Onshape and Creo points to a broader pattern: AI for product development is moving from guidance toward integrated automation, with product data and reusable engineering knowledge as the foundation.
Your membership also unlocks:
