Design Client/Server Codebase Sync with a Hash Tree
This is where the Merkle tree screen goes next. Candidates describe the screen continuing into a client and server sync as its second part, and in August 2026 a Cursor technical screen asked for the sync as a design of its own. A 2025 online test paired the tree with client-server tasks too. The wording varies, so expect the shape rather than a script.
The shape is worth preparing for anyway, because it's the Merkle tree screen grown up. In the Merkle tree coding lesson you build the tree and diff two of them in memory, on one machine. That lesson already covers the hashing rule and the recursive diff, so this one doesn't repeat them. Here the two trees live on different machines, with a slow network between them. A sync can stop halfway, and the server keeps publishing new versions while clients are mid-sync. Those three facts drive the design. The diff now costs round trips, so we want as few as we can get. A half-finished sync leaves the client in a state neither version describes. And a moving target means every sync has to pick one version and stick to it.
CursorPremium lesson · System design round
The Cursor design, worked end to end.
- Requirements and scopeSystem design
- The API and data model
- The design, drawn out1 diagram
- 5 deep divesWith the tradeoffs
$49 a year · all 107 companies · instant access
20,000+ candidate reports
Cursor interview guideKnow what Cursor asks before you walk in.
9 coding · 4 system design · 16 behavioral questions
Get their questions first.
InterviewLegend is independent and not affiliated with or endorsed by any company it covers.