Back to SpeedPy

Your first project

Get SpeedPy running on your computer

You will choose an AI coding agent, prepare your computer, copy SpeedPy, and let the agent finish the setup. You do not need to understand Docker or write the application by hand.

Showing instructions for

We selected this from your browser. Change it in any operating-system selector below and every section will follow.

  1. 01

    Choose your agent

  2. 02

    Prepare the terminal

  3. 03

    Install Docker

  4. 04

    Copy SpeedPy

  5. 05

    Ask the agent

Step 1 · Choose one

Pick the AI agent you will build with

An AI coding agent can read the whole project, change files, run commands, and check its work. You only need one. If you already use one of these, keep using it.

Each link goes to the product’s official setup page. Install the agent, sign in, and come back here. SpeedPy includes the same starter skill for all four.

Claude Code

Terminal, editor, or desktop

Codex

App, terminal, or editor

Cursor

Code editor with an agent

Kimi Code

Terminal or editor

Using Windows? Choose your agent now. If you choose a terminal agent such as Claude Code or Kimi Code, install it from the Ubuntu terminal after the next step.

Step 2 · Prepare the terminal

Give your agent a place to work

The terminal is a window where you or your agent can give the computer precise instructions. Windows needs one extra preparation step.

Install Ubuntu inside Windows

SpeedPy runs in a Linux environment. Windows Subsystem for Linux (WSL 2) provides that environment without replacing Windows.

1. Open PowerShell as Administrator and run:

wsl --install

2. Restart the computer when Windows asks. Then open Ubuntu from the Start menu and create the username and password it requests.

3. In the new Ubuntu terminal, install Git:

sudo apt update
sudo apt install -y git
Open Microsoft’s WSL guide

Use the Terminal app

Open Spotlight with ⌘ Space, type Terminal, and open it. Run this check:

git --version

If macOS offers to install developer tools, accept the installation. That supplies Git, which keeps track of your project.

Open your terminal and check Git

Most Linux systems already include Git. Check before installing anything:

git --version

If the command is missing, use your distribution’s software manager to install Git. Your AI agent can identify the correct command for Ubuntu, Debian, Fedora, or another distribution.

Step 3 · Install Docker

Install the box SpeedPy runs inside

Docker gives SpeedPy the same predictable environment on every computer. You do not need to learn how Docker works before you start.

Install Docker Desktop for Windows

  1. 1. Download and install Docker Desktop using the WSL 2 backend.
  2. 2. Open Docker Desktop. In Settings → Resources → WSL Integration, enable your Ubuntu distribution.
  3. 3. Return to the Ubuntu terminal—not PowerShell—for the rest of this guide.
Install Docker Desktop

Install Docker Desktop for Mac

Use Docker’s installer for your Mac, then open Docker Desktop from Applications. Wait until it reports that Docker is running.

Install Docker Desktop

Install Docker Engine and Docker Compose

Choose your Linux distribution in Docker’s guide and install both Docker Engine and the Compose plugin. The commands differ between distributions, so use the matching official page.

Open Docker’s Linux guide

Check that Docker is ready in your terminal:

docker info
docker compose version

If both commands finish without an error, continue. If not, leave Docker Desktop open and ask your coding agent to diagnose the message.

Step 4 · Copy SpeedPy

Create your project folder

“Clone” simply means copying the SpeedPy project from GitHub to your computer. Replace my-product with a short name for your idea.

Run in your Ubuntu terminal
mkdir -p ~/projects
cd ~/projects
git clone https://github.com/speedpy/speedpy.git my-product
cd my-product
Keep this folder inside Ubuntu. The ~/projects location avoids the slower and less reliable Windows-mounted folders under /mnt/c.
Run in Terminal
mkdir -p ~/projects
cd ~/projects
git clone https://github.com/speedpy/speedpy.git my-product
cd my-product
Run in your terminal
mkdir -p ~/projects
cd ~/projects
git clone https://github.com/speedpy/speedpy.git my-product
cd my-product

Step 5 · Hand it to your agent

Let the SpeedPy starter skill finish the setup

Open the my-product folder in the coding agent you chose. SpeedPy includes a skill that checks your exact computer, explains anything missing, runs init-docker.sh, and verifies the site.

Claude Code or Kimi Code

Stay in the project terminal and run claude or kimi.

Codex or Cursor

Open the app, choose Open folder, and select projects/my-product.

Tell your coding agent

Use the start-speedpy skill to check this computer and start SpeedPy with Docker.

1

The agent checks everything

2

It runs the initializer

✓

It gives you the local link

What happens next

Describe the business you want to build

Once SpeedPy is running, your agent can help turn the idea into a clear specification, ask about missing details, and build the product on top of the working foundation.

Explore the SpeedPy documentation