How the DoesItHold training loop works

1. Generate a challenge

Pick a language and start a challenge. On the Free plan you get one freshly generated challenge each day, plus curated public challenges you can repeat at any time.

2. Read and diagnose

Each challenge is a short snippet with a real defect. Read it as you would a pull request: form a hypothesis about what breaks and why before you touch anything.

3. Propose a fix

Edit the code and submit. If you get stuck, request a hint; hints escalate from where to look toward what to change, so you keep learning instead of stalling.

4. Read the feedback

Feedback tells you whether the fix holds and explains the cause. The evaluator also catches fixes that look plausible but leave the defect in place, which is the most useful signal for learning.

5. Repeat at an adapting difficulty

Difficulty adjusts to your recent results. The aim is steady, useful practice; progress reflects what you have done in the app, not a claim about outcomes elsewhere.

Related pages