# How to Install text-to-CAD: Skills CLI, Plugin, and Development Setup

> Easily install text-to-CAD with the Skills CLI or provider-specific plugins for Codex and Claude Code agents. Get started with text-to-CAD today!

- Repository: [earthtojake/text-to-cad](https://github.com/earthtojake/text-to-cad)
- Tags: getting-started
- Published: 2026-08-02

---

**Install text-to-CAD using `npx skills install earthtojake/text-to-cad` for the standard Skills CLI method, or use provider-specific plugins for Codex (`codex plugin add cad@text-to-cad`) and Claude Code (`claude plugin install cad@text-to-cad`) agents.**

The `text-to-cad` repository by earthtojake provides reusable agent skills for CAD, robotics, and hardware design. Because the project ships pre-generated skill output on the `main` branch, you can install the library through multiple pathways depending on your agent runtime. This guide covers the Skills CLI method, alternative plugin-based installations, and the development workflow for contributors.

## Install text-to-CAD via the Skills CLI (Recommended)

The **Skills CLI** is the preferred production method for installing text-to-cAD. This approach works with any agent that supports the Skills framework and automatically registers individual skills—including CAD, URDF, and G-code handlers—in your agent’s skill registry.

According to the repository's [`README.md`](https://github.com/earthtojake/text-to-cad/blob/main/README.md) (lines 69-71), run the following command:

```bash
npx skills install earthtojake/text-to-cad

```

This command pulls the `text-to-cad` package and configures each skill for immediate use.

### Verify the Installation

After installation, confirm that the CAD skill is available in your environment:

```bash
skills list | grep cad

```

## Alternative Installation Methods

If you are using Codex or Claude Code agents, you can install text-to-CAD through provider-native plugins that expose the same skill set without using the Skills CLI directly.

### Codex Plugin Installation

For Codex agents version 0.142.0 or higher, use the marketplace commands documented in [`README.md`](https://github.com/earthtojake/text-to-cad/blob/main/README.md) (lines 78-84):

```bash
codex plugin marketplace add earthtojake/text-to-cad
codex plugin add cad@text-to-cad
codex restart

```

The `codex restart` command ensures the new skills are loaded into the runtime.

### Claude Code Plugin Installation

For Claude Code environments, the installation process follows a similar pattern, as specified in [`README.md`](https://github.com/earthtojake/text-to-cad/blob/main/README.md) (lines 90-94):

```bash
claude plugin marketplace add earthtojake/text-to-cad
claude plugin install cad@text-to-cad

```

## Using text-to-CAD After Installation

Once installed, you can invoke CAD generation capabilities programmatically. The skills expose Python APIs for creating STEP and GLB files through the shared package implemented in [`packages/cadpy/README.md`](https://github.com/earthtojake/text-to-cad/blob/main/packages/cadpy/README.md).

### Generate a STEP Model

The following example demonstrates generating a mechanical part using the `CadSkill` class:

```python
from skills.cad import CadSkill

cad = CadSkill()
step_file = cad.create(
    prompt="Create a 100 mm cube with a 10 mm hole through the center",
    export="step"
)

print(f"Generated STEP file saved at: {step_file}")

```

## Development Installation

When contributing to text-to-cad or testing locally, you must work on the `develop` branch rather than `main`. The [`CONTRIBUTING.md`](https://github.com/earthtojake/text-to-cad/blob/main/CONTRIBUTING.md) file outlines the required workflow for linking skill sources and running the test suite.

### Setup Development Symlinks

The repository uses a symlink layout for local development. Run the setup script to create the necessary links:

```bash
scripts/dev/setup-symlinks.sh --check

```

### Run Tests

After making changes, validate your work by running the Python-specific test suite:

```bash
./.venv/bin/python -m unittest tests/python/*_test.py

```

This ensures that modifications to skill logic in [`skills/cad/SKILL.md`](https://github.com/earthtojake/text-to-cad/blob/main/skills/cad/SKILL.md) or related packages function correctly before submitting a pull request.

## Summary

- **Skills CLI** (`npx skills install earthtojake/text-to-cad`) is the recommended method for production environments and works across any Skills-compatible agent.
- **Codex and Claude Code plugins** provide native integration for those specific agent runtimes using marketplace commands.
- **Verification** via `skills list | grep cad` confirms successful registration of CAD capabilities in your environment.
- **Development workflows** require the `develop` branch and symlink setup via [`scripts/dev/setup-symlinks.sh`](https://github.com/earthtojake/text-to-cad/blob/main/scripts/dev/setup-symlinks.sh) before running tests.

## Frequently Asked Questions

### What is the difference between Skills CLI and plugin installation?

The **Skills CLI** method is the universal approach that works across any agent supporting the Skills framework, while plugin installations are provider-specific optimizations for Codex and Claude Code that may offer tighter integration with those environments. Both methods provide access to the same underlying skills defined in the repository.

### Which branch should I use for development?

Always use the **`develop`** branch when contributing to the earthtojake/text-to-cad repository. The `main` branch contains pre-generated skill output intended for end-users, whereas `develop` contains the source files and symlink layout required for editing and testing changes.

### How do I verify that text-to-CAD skills are properly registered?

Run `skills list | grep cad` after installation. If the CAD skill appears in the output, the text-to-CAD package is correctly installed and available to your agent runtime. You can also test functionality by importing `CadSkill` from `skills.cad` in a Python session.

### What dependencies are required for the development workflow?

You need **Python** configured in a virtual environment (`.venv`) and must execute [`scripts/dev/setup-symlinks.sh`](https://github.com/earthtojake/text-to-cad/blob/main/scripts/dev/setup-symlinks.sh) to create the required development symlink layout before running tests. The test suite relies on `unittest` and validates changes to the Python implementation files.