Was ist ein MCP-Server und in welcher Beziehung steht er zu CAD?

MCP may be one of the most important technologies you have never heard of.

It stands for Model Context Protocol. In plain English, MCP is an open standard that allows an AI application to connect to an external system and use its tools.

A simple way to think about it:

If APIs are how software talks to applications, MCP is how AI talks to applications.

Without an MCP, an AI model can explain what you should do. With MCP, it can access the right tools, perform authorized actions, examine the results, and improve its work.

That is a major difference.

Why Is MCP Important to CAD?

MCP is how AI will communicate with CAD applications.

Tools such as Codex, Claude Code, and GitHub Copilot can act as MCP clients. The CAD system provides an MCP server that defines what the AI can access and what actions it can perform.

The AI client provides the intelligence.

The MCP server provides the specialized CAD tools.

This can allow an AI agent to create geometry, modify parameters, inspect a model, export information, or automate a multistep workflow.

But connecting AI to CAD does not automatically make the AI a capable engineer.

The Connection Is Only the First Step

Most CAD systems were designed around human interaction with predefined commands.

An MCP server may allow AI to use those commands, but the AI is still operating within the existing toolset. To create something complex, it may need to assemble long sequences of sketches, constraints, features, and edits.

That introduces several challenges:

• Long command sequences can be fragile
• Repeated operations consume time and AI tokens
• Geometry may be created without robust design intent
• Complex models can become difficult for humans to understand and edit
• The AI may need to become involved again every time a variation is required
• The result may be a one-time model rather than a reusable engineering capability

MCP solves the communication problem.

READ: How to Get Started with the Onshape FeatureScript MCP Server

It does not, by itself, solve the deeper challenge of creating predictable, editable, and reusable engineering tools.

This is where Onshape takes MCP in a new direction.

Onshape’s Unique Advantage: FeatureScript

FeatureScript is Onshape’s built-in programming language for creating parametric CAD features.

It is not a separate macro system or an external automation layer. It is the same language Onshape uses to create many of its native modeling tools.

READ: What Is FeatureScript? How It Differs from CAD Macros and APIs

A FeatureScript custom feature can:

• Create complex parametric geometry
• Capture design intent
• Generate a purpose-built user interface
• Regenerate as the design changes
• Appear directly in the Onshape toolbar
• Be shared across an engineering organization

FeatureScript has always made Onshape highly customizable. The challenge was that creating sophisticated custom features required programming expertise.

The new FeatureScript MCP Server from Onshape Labs dramatically lowers that barrier.

AI Can Now Build New CAD Tools

Connect the FeatureScript MCP Server to an AI client such as Codex, Claude Code, or GitHub Copilot, and an engineer can describe the capability they need in plain language.

The AI can then:

• Reference FeatureScript documentation
• Generate the code
• Insert it into an Onshape document
• Run and test the feature
• Identify errors
• Debug the code
• Refine the feature until it works

The AI is not limited to operating the CAD commands that already exist.

It can help create entirely new CAD capabilities.

That is an important new direction for AI and CAD.

This Is Text to Code to CAD

Imagine asking an AI to create a custom feature that adds a manufacturable garden-hose thread to a selected cylindrical face.

The feature needs to determine whether the thread is internal or external, apply the correct geometry, allow the engineer to control its length, and adapt when the underlying part changes.

A traditional AI-to-CAD workflow might attempt to construct the thread through a long series of individual modeling actions. That may successfully produce one part, but it does not necessarily yield a robust, reusable tool.

With FeatureScript MCP, the AI creates the code that defines how the feature works.

The result is not one threaded part. It is a native, editable, parametric thread feature that can be added to the Onshape toolbar and applied to future designs.

The workflow becomes text-to-code-to-CAD – the prompt defines the need, the AI helps create the code, the code creates predictable CAD geometry, and the engineer remains in control.

Build Once and Use Repeatedly

This approach helps overcome another challenge with AI-driven CAD: repeatedly asking the AI to perform the same work

Once a custom feature has been created and validated, the AI is no longer required each time the feature is used. The FeatureScript runs inside Onshape like any other parametric feature.

The tool can be:

• Reviewed and refined by engineers
• Shared across the organization
• Applied consistently by every designer
• Updated as company standards evolve
• Reused without repeated prompts or AI token consumption

AI helps build the tool once. Engineers can use it thousands of times.

A CAD System Tailored to Your Company

Every company has its own products, standards, manufacturing processes, and engineering practices.

Yet most engineering teams work with essentially the same general-purpose CAD commands.

FeatureScript MCP creates the potential for a production CAD system tailored to the way a company actually designs.

An organization could use AI to help build:

• Company-specific mounting features
• Hydraulic ports based on internal standards
• Configurable components and product families
• Specialized manufacturing geometry
• Automated engineering calculations
• Design-rule and validation checks
• Tools that turn repetitive processes into a single command

This is more than CAD automation. It is a way to codify engineering knowledge and make it reusable throughout the organization.

The engineers who understand the products and processes can describe what the tool needs to do. AI can help translate that expertise into FeatureScript. Engineers can then inspect, test, refine, and share the resulting capability.

Why This Matters

MCP is how AI will communicate with CAD applications.

That connection will enable AI agents to perform useful modeling and engineering tasks across many CAD platforms. But connectivity alone is not the destination.

The real opportunity is giving AI the ability to help engineers expand the CAD system itself.

Onshape’s unique advantage is the combination of MCP and FeatureScript. Instead of limiting AI to existing commands or one-time geometry, Onshape enables AI to create native, parametric, reusable engineering tools.

Only Onshape combines cloud-native CAD and PDM with AI-powered customization, enabling companies to build engineering capabilities tailored to their products and processes.

MCP connects AI to CAD. FeatureScript MCP turns that connection into a CAD customization engine.