Your code passes. Does your thinking?
Trace writes the problem, checks your output against real test cases, then grades the way you solved it — the reasoning, the complexity, the tradeoffs you missed. Stuck? Hints unlock one step at a time.
Three steps, every single time
No dashboards to configure. No curated lists to grind through. The problem exists only because you showed up to solve it.
- 01
Problem is written for you
Tell Trace what you want to get better at — graphs, DP, concurrency — and it drafts a problem with constraints, examples and a hidden test suite.
- 02
You solve it, not guess it
Write the solution in a real editor. TypeScript and Python run today, and every result is checked against generated test cases.
- 03
Your approach gets graded
Then the interesting part: AI reads your code and explains what your solution reveals about your thinking, and what to fix next.
Output is checked by tests. Approach is checked by AI.
We split grading on purpose. Machine-checkable results stay deterministic — no hallucinated pass, no generous judge. The parts a test cannot see are where the AI earns its place.
Value checking
Deterministic. 14 cases, 5 visible and 9 hidden, exact output comparison, time and memory limits.
AI review
Reads your code for complexity, edge cases, naming and the tradeoffs you skipped.
Nudges that keep you thinking
Hints never hand you the code. Each one is a single step forward — a question, a constraint to re-read, a pattern to recognise — generated from your own attempt.
- 1
You are re-scanning the array for every element. What would let you answer “have I seen this value?” without looking again?
- 2
A hash map turns a second loop into a constant-time lookup. Which value should be the key, and which the value?
- 3
Your map stores values seen so far. At each index, the partner you need is `target − nums[i]`. Check before you insert.
- 4
Edge case: duplicate values such as [3,3]. Does your check run before or after the insert?
- 5
Full walkthrough with the optimal solution and reasoning
Structured, not spoon-fed
Five rungs from “re-read the constraint” to a full walkthrough. You choose when to stop climbing.
Scored transparently
Every hint you open is recorded in your review, so a perfect score means you got there unaided.
Learns your gaps
Hints cluster around the patterns you keep missing — sliding window, monotonic stacks, off-by-one bounds.
Solve one problem tonight. Read what it says about you.
Five AI-generated problems a week, unlimited AI review, and the full hint ladder. Upgrade only when you want deeper roadmaps.