Launching October 21.
bug.dr

AI can write the code. Can you fix it when it breaks?

bug.dr drops you into a real, broken production codebase with an AI assistant in your editor. You investigate, fix and verify. big.dr scores how well you did it.

Join the waitlistSee how it works
Launching October 21Runs in your browserAI included
  • Bugs you could face at work
  • Code, terminal and AI in one place
  • Clear checks for every fix

The job changed. The coding test barely did.

AI can suggest a fix in seconds. Knowing whether it's the right fix takes practice. On bug.dr, you work on bugs, making sense of the code, and checking that your changes solve the problem without creating another one.

The task
Solve a standalone puzzle.
Investigate a bug in an existing codebase.
Using AI
Often restricted.
Part of the workspace.
The work
Write an answer from scratch.
Read code, diagnose the issue, make changes, and test them.
The checks
Does the answer match the expected output?
Does the bug stay fixed, and does the rest still work?
What you can show
Your answer.
Your fix and how you got there.

How it works

  1. 1

    Get the context

    Read the bug report and explore the codebase.

    You'll know what the app is supposed to do and where it's failing. Finding the cause is your job. The timer starts when you start the case.

  2. 2

    Work through the fix

    Use the editor, terminal, and AI assistant in your browser.

    Follow a stack trace, ask questions, try a change, and run the code. Your prompts, token usage, and test runs are recorded along the way.

  3. 3

    Check your work

    Run the checks when you're ready.

    Your fix needs to solve the problem without breaking the rest of the app. Once you pass, your result is saved. You can't replay the same case to replace it with a better score.

Payment retries problem page showing the assignment, incident log and acceptance checks.
bug.dr profile page showing progress and solved cases

Get the link in your inbox the day we open.

Join the waitlist

Pick a role.
Build your way there.

Stop guessing what to practice. Follow a path built around the work your target role actually involves, one well-solved problem at a time.

EXPLORE A PATH

AI Engineer

Models are only part of the job.
Practice the backend, data, and interfaces too.

STAGE 01 / EASY

Small bugs. Strong foundations.

Trace a failure, make a focused change, and prove it works.

Pass the checks. Meet the efficiency benchmark. Move forward.
WHAT YOU'LL WORK ON Illustrative examples
AI / ML

A prompt that exceeds the context window

Backend

An API response the client cannot parse

Database

A vector search returning duplicate results

You fixed it. Now learn from it.

After each solve, get a debrief based on your session. See how many prompts, tokens, and test runs you used, how you compare with other people on that case, and what you could try differently next time.

Session debriefExample session
Payment retries7 / 7 checks passed
You

It passes. What could I have done better?

bug.dr coach

You fixed it. Your change passed every acceptance check. The next thing to work on is how you get to the fix.

This sessionYouTop performers
Prompts124
Tokens3,8401,200
First-run passNo68%

You ran tests 8 times before your first passing run. Next time, spend a little longer on the logs before asking AI for a change.

Try this on your next case

  1. 1Read the full stack trace.
  2. 2Find the likely failure point.
  3. 3Ask for one targeted change, then test it.

A working fix is the start. A repeatable approach is what you take into the next problem.

Illustrative feedback and benchmark data.

Generated automatically after every solve. Free tier includes feedback on Easy solves.

Practice the work you want to do.

Choose your role and start with cases that fit your experience. Build confidence with easier fixes, then take on problems that need a deeper investigation.

BEBackend
A connection pool that runs out under load.
FEFrontend
A page that keeps rendering until it slows to a crawl.
FSFull-stack
Two users manage to book the same slot.
DBDatabase
An orders dashboard slowed down by repeated queries.
AIAI / LLM
A summarizer sending more context than the model can handle.
DODevOps / SRE
A service that keeps running out of memory.
bug.dr profile page showing points, current level, current streak, a twelve-month activity grid and a list of solved problems.

Something's broken. Pick your case.

Duplicate charges. Slow queries. Failed deployments. Browse by role, topic, and difficulty to find something worth working through. Every case includes a brief and the checks you need to pass, so you know what a working fix needs to do before you start.

Keep your next challenge in sight.

Find your next case, keep track of your progress, or test yourself against the same challenge as other engineers.

Contests

A shared challenge. Your own approach.

Take on the same case. Daily, weekly, and monthly contests give you a shared challenge. See how your approach compares when everyone starts with the same problem.

Contests page showing live daily, weekly and monthly incidents with countdowns and a contest history table.

Start free. Go deeper when you're ready.

Get started with bug.dr and see how you work through a case.

Free
€0/month

Included

  • All Easy problems
  • Basic AI assistant
  • Efficiency scoring
  • Public leaderboard
  • AI feedback on Easy solves

Not included

  • Career paths
  • Medium, Hard and Get a job problems
  • Powerful AI model
Join waitlist
Pro
Save €57
€8.25/month
Billed €99 yearly

Included

  • Everything in Free
  • Career paths with efficiency gating
  • AI coaching after every solve
  • Powerful AI model included
  • Connect your own API key
Join waitlist

Payments open after launch.

Be the first to pick a problem when the library opens.

Join the waitlist

Your first case is coming.

We're opening bug.dr on October 21. Leave your email and we'll send you a reminder the night before and a link when it's live.

A few things you might be wondering.

Yes. There's an AI assistant in the workspace. Use it to explore the code, ask questions, and work on your fix. You still need to understand the changes and check that they work.

They're practice codebases built around bugs and failures you could encounter at work. You get the context of a production incident in an environment where you can investigate and make changes.

The library covers cases using Node.js, TypeScript, Python, Go, React, Next.js, PostgreSQL, Redis, JWT, and WebSockets. Check each case for its stack before you start.

No. The editor, terminal, AI assistant, and tests run in your browser.

A working, reliable fix matters most. We also look at code quality, how you test your changes, time, and AI usage. Using fewer tokens or writing fewer lines doesn't automatically make a fix better.

Your first successful result is final. You can learn from it and take that experience into your next case.

Pro adds a personalized AI debrief after each solve and structured career paths. It costs €12.99 a month or €99 a year.

No platform can promise that. A path helps you build experience and see how your work compares with its benchmarks. Hiring decisions still belong to employers.

October 21, 2026. Join the waitlist and we'll send you the link when we open.

Yes. Join the waitlist and we'll reach out about using bug.dr with your team.