How Turing works
Timed practice on deliberately broken repositories, with tests, your own checks and an assistant that helps you investigate without writing the fix.
- 1
Pick a broken repository
Each challenge is a small project with deliberately faulty behaviour, written in the language and framework you want to practise.
- 2
Start a timed attempt
Your workspace opens with the problem, the requirements, the file tree and the supplied checks. The timer is kept on the server.
- 3
Investigate and repair
Read the code, run the supplied checks and write your own tests in the console. Every save becomes a revision you can browse later.
- 4
Ask the assistant
The assistant explains code, errors and language features using your saved revision. It helps you investigate; it does not write the fix.
- 5
Submit and review
Submission runs the supplied and hidden checks on your final revision and opens a debrief with your score, teaching notes and an AI-usage review.
Workspace rules · enforced by the server
- Editable · Source files of the challenge language
- Locked · Supplied tests in supplied-tests/, configuration and build files
- Checks · Run against saved revisions only
Privacy
Your attempts, saved code, assistant conversations and support messages are stored for review. When you use the assistant, your question and the relevant saved code are shared with the AI service. Avoid putting secrets or personal information in code or messages.
Practice terms
Exercises are provided for practice. Visible checks describe the supplied scenarios; passing them is not a certification or a hiring outcome. Timers continue after a tab is closed, and checks use your saved revision.
Contact support
Use Support to report a bug, a wrong test or an assistant issue. Replies and the current status appear in the same thread.