BugForge
BugForge / 00Runs in your browser

Break it.
Fix it.
Build it.

Real codebases with failing tests, an endless bug-injection engine, build-from-scratch labs, spot-the-bug drills and DSA. Free, and it all runs in your browser.

debug labs
23
build labs
9
bug drills
65
DSA problems
30
Your forge · start today
0hunt streak
0hunts cleared
0labs solved
0drills correct
0DSA solved
Stats
fig. 01poke it
01 / Repository debuggingBreak it

Debug labs

Every lab ships a real codebase, a bug ticket and a red test suite. Read the first failure, form one hypothesis, change one thing.

Labs
23
Stack
Node · Python
Bugs
~3 per lab
  1. 1Read the ticket→
  2. 2Run the tests→
  3. 3Trace→
  4. 4Fix→
  5. 5Verify
Open a lab
fig. 02poke it
02 / Endless repsFix it

Bug Hunt

A mutation engine plants fresh, verified regressions in working code. Clear them against the clock until the patterns are reflexes.

Bugs
Endless
Mode
Timed
Track
Streaks
  1. 1Get a mutant→
  2. 2Spot the diff→
  3. 3Patch→
  4. 4Next
Start hunting
fig. 03poke it
03 / ImplementationBuild it

Build labs

A spec, some stubs and the tests. Brew rate limiters, queues, caches and auth from scratch until every check turns green.

Labs
9
Start
Spec + stubs
Done
All green
  1. 1Read the spec→
  2. 2Fill the TODOs→
  3. 3Run→
  4. 4Ship
Build something
BreakFixBuildRepeat
04–06 / More modes

Sharpen the edges.

Speed reading, patterns and the concepts underneath. Point at a sprite to poke it.

07 / Field guide1 bug × 6 renders

Same bug,
six disguises.

A regression looks different in every codebase. Learn its shape, not its outfit.

01Original
02Mosaic
03Rounded
04Halftone
05Cross-stitch
06Dither
08 / The assessment, decoded

What the online assessment looks like.

Formats vary by level and batch (candidate reports, not an official spec), but these blocks show up again and again.

40–90 min · 1–2 problems

DSA coding

LeetCode-medium style in a HackerRank-like editor. Arrays, sliding window, hashing, heaps, BFS/DFS, greedy, DP. Hidden tests hit edge cases and large inputs.

≈ 60 min · ~3 bugs

AI-assisted debugging

A real codebase, a ticket, failing tests and an AI assistant. Reproduce, trace, hypothesize, fix, verify. Exactly what the Debug Labs train.

behaviour signals

What graders love

Small targeted diffs, reading the failing assertion first, not editing tests, checking the AI's suggestions, clean edge-case handling.

09 / Daily loop≈ 108 min

About two hours a day.

Repetition is the whole trick. Run this loop daily and bug patterns become reflexes.

  1. 011 Bug Hunt warm-up10 min
  2. 0210 Spot-the-Bug speed reading8 min
  3. 031 Debug Lab OA simulation30 min
  4. 041 Build Lab implementation30 min
  5. 051 DSA problem pattern practice30 min
10 / Open source

Free forever, MIT licensed. If it helps you prep, leave a star.

New labs, lessons and drills are welcome as pull requests: every one is verified at build time.

BugForge · MITRuns locally in Web Workers · progress stays in your browser