Your machine. Your workspace.

Set up your computer to run code

crackedai-connect is the program that runs your exercises. Install it on your computer, start it in a terminal, then return to your exercise. This page checks whether PyTorch is ready on this device.

Install


                pipx install crackedai-connect

Requires Python 3.12 or newer and pipx. Install Python and pipx first, or use the uv option below.

Other installers, Python versions, and upgrades

Have uv installed? uv tool install crackedai-connect creates an isolated environment and can download a suitable Python if needed. See the uv installation guide if you need uv first. Use one installer for this runtime to avoid conflicting commands.

pipx installs with its default Python. If that is older than 3.12, add --python followed by the path to a Python 3.12 or newer executable. See the pipx installation guide for pipx itself.

Use the latest crackedai-connect release. Stop the running copy with Ctrl+C before updating.

Installed with pipx? Run pipx upgrade crackedai-connect, then start crackedai-connect again.

Moving a pipx installation from an older Python? Install Python 3.12 or newer, then run pipx reinstall --python PATH crackedai-connect, with PATH pointing at that interpreter. This recreates the runtime environment; framework packages may need to be downloaded again when you restart it.

Installed with uv? Run uv tool upgrade crackedai-connect, then restart the runtime.

Managing your own environment? Use Python 3.12 or newer, activate the environment, then run python -m pip install --upgrade crackedai-connect. Avoid installing into an operating-system-managed Python environment.

Run

crackedai-connect

For PyTorch

Choose PyTorch when the runtime asks which modes to set up. It downloads a build matched to your hardware, which can take several minutes. CPU-only machines work too.

Add or change modes

Stop the runtime with Ctrl+C, then run crackedai-connect --reconfigure. Choose every mode you want to keep, including the new one.

Or provide the full list directly, for example crackedai-connect --toolchains pytorch,jax,rust. This replaces your saved selection; it does not add to it. Available names are pytorch, jax, python, rust, and cpp.

Solve

Keep the runtime running in your terminal. Return to your exercise when the status above says your mode is ready.

Browse problems

Compare all mode requirements

PyTorch and JAX: The installer checks your hardware to select framework packages. GPU support depends on your platform, drivers and framework version. CPU execution is also supported. Framework downloads can take several minutes and use several gigabytes of disk space.
Python: mypy checks type annotations; test cases check results. The runtime’s Python interpreter executes your code.
Rust and C++: Install a compiler separately, along with clippy or clang-tidy for submission checks. The FAQ below explains how to install these tools.

FAQ

I already have PyTorch or JAX installed

The runtime uses packages from the Python environment where it is installed. A pipx install has its own isolated environment, so it cannot reuse packages from a different virtualenv, conda, or Pixi environment.

Use an environment I already manage

Activate that environment in your terminal.
Install the runtime there with python -m pip install crackedai-connect. The environment needs Python 3.12 or newer.
Start it with crackedai-connect --skip-setup to use your existing packages without the automatic toolchain installer.

You manage any missing dependencies yourself. Check the connection status above for your selected mode; some JAX exercises also need Flax and Optax. To let the runtime manage setup again, restart without --skip-setup.

Check connection status ↑

On a phone or tablet? Use a computer you already own

The runtime is a desktop program, so there is nothing to install here. Instead, link a computer that has it running and your phone can send code there and receive the results. The local check at the top of this page can keep waiting on your phone even when your linked computer is ready.

On your computer, install the runtime using the steps above. Sign in to Settings with the account you’ll use on your phone, then choose Link a machine under Linked machines.
Run crackedai-connect --link CODE on the computer, using the code from Settings. Then start crackedai-connect and leave it running.
On your phone, sign in to the same account and open your exercise. Run and Submit will name the linked computer.

Browse problems

Keep the linked computer awake with the runtime running. It does not need a browser open.

Safari connection help. If detection keeps waiting after you start the runtime, try opening crackedaicode.com in Chrome, Edge or Firefox. You can keep the same runtime running; there is no need to reinstall it.

It says "Listening" but this site can't see it

Allow browser access. If your browser asks to connect to your local network, choose Allow. If access is blocked, open the site’s permissions beside the address bar and allow local network access.
Update and restart the runtime. With pipx, run pipx upgrade crackedai-connect. For uv or your own environment, open "Other installers, Python versions, and upgrades" in step 1. The runtime needs Python 3.12 or newer. Restart crackedai-connect after the update.
Check your mode is ready. Return to the connection status and choose Check again. Follow any missing-tool instructions it reports.

Back to connection status ↑

Still not connecting?

On Safari, try Chrome or Edge if it still won’t connect. Safari can be stricter about a secure page reaching a program on your own machine.

Check the runtime directly: open http://127.0.0.1:52341/health in a tab. A JSON response confirms it is reachable from that tab. It does not confirm that this site can connect or that your selected mode is ready; use the connection status for that.

Where does my code run?

It runs in crackedai-connect on your computer or a machine you have linked. CrackedAI’s web server does not execute it. The editor sends your code to the site as you type so it can prepare Run and Submit requests.

When you are signed in, the site saves drafts and accepted solutions so you can return to your work and track your progress. Privacy lists what is kept.

What does Python mode need?

Python-mode submissions are graded on mypy --strict as well as on the tests, so the runtime needs mypy. Choose Python when it first sets up, or follow Add or change modes to include it in your saved selection.

What does Rust mode need?

A Rust toolchain on your PATH. rustup.rs installs one. Rust code is compiled locally, so without rustc it can't run at all.

Submissions also need a clean Clippy check, with warnings treated as errors. Clippy is included in rustup's default installation. If you chose a minimal installation or the runtime reports it missing, run rustup component add clippy, then restart the runtime.

What does C++ mode need?

The runtime looks for clang++ or g++ on your PATH. Submissions also need clang-tidy. Install the tools in the same environment where you run crackedai-connect.

macOS: xcode-select --install provides the compiler. Install clang-tidy with brew install llvm, then follow the PATH instructions below if it is not detected.
Debian/Ubuntu: Install both tools with sudo apt install g++ clang-tidy.
Windows: Use Ubuntu in Windows Subsystem for Linux (WSL) and follow the Debian/Ubuntu step. Install and run crackedai-connect inside that same WSL environment. Visual Studio's MSVC compiler (cl.exe) is not detected by the runtime.

Installed LLVM on macOS, but clang-tidy is still missing?

Homebrew keeps LLVM's tools in a separate folder. After installing LLVM, stop the runtime with Ctrl+C, then run this in the same terminal:


                  export PATH="$PATH:$(brew --prefix llvm)/bin"

Check clang-tidy --version, then start crackedai-connect again from that terminal. This change applies to this terminal session; repeat it when opening a new one.

Can I use a virtual environment?

Yes. Use a venv, conda or Pixi environment with Python 3.12 or newer. The runtime uses that environment for its Python packages. Follow the “Use an environment I already manage” instructions above to keep control of dependency installation.