UOJ Prime Round 2026 will be held at 10:00 AM on Saturday, August 8 (UTC +8)! The contest will last for 672 hours (28 days).
UOJ Challenge #1 will be held at 10:00 AM on Sunday, August 9 (UTC +8)! The contest will also last for 672 hours (28 days).
UOJ Prime Round has already been mentioned in a previous blog post. It will bring together problems from on-site contests over the past year that were solved by almost nobody during the contests and have received relatively few post-contest submissions.
As for UOJ Challenge, Gellyfish happened to have an interesting problem that did not seem to fit particularly well anywhere else, so we thought we might as well release it alongside UOJ Prime Round for everyone to try! It will be a fairly challenging problem combining Algorithms and Heuristics, and we hope everyone has fun with it (?).
UOJ Challenge #1 is rated, with rating changes scaled to $$$\frac{1}{125}$$$ of those in a normal rated contest.
The winner of UOJ Challenge #1 will receive a piece of UOJ merchandise! You may choose either a UOJ pillow or a UOJ power bank.
To keep things mysterious, that is all we are going to reveal for now! We wish all participants an enjoyable experience in these two special contests!
Problem setters: Gellyfish, aberter0x3f.
The following is an advertisement for Hull from aberter0x3f:
Hull is a next-generation competitive programming problem-setting, analysis, and packaging framework built on Nix and WASM. It integrates problem configuration, programs, and statement documents into a modern project that is reproducible, testable, and exportable with a single command.
Hull uses Nix + WASM to achieve fully reproducible environments and deterministic judging, eliminating differences between environments and random fluctuations. Its high-precision sandbox supports time and memory measurements accurate down to 1 tick / 1 Byte. It also provides built-in support for multi-process interactive communication, a virtual file system, and automatic deadlock detection, so you no longer have to worry about pipes hanging in interactive problems!
Write once, export everywhere: there is no longer any need to manually modify problem.conf or write complicated ZIP packaging scripts. Hull can export the same problem set into:
- Dedicated judging bundles for UOJ / Hydro / Lemon, with support for custom Supervisors and static dependency packaging
- Platform-compatible packages for Luogu / CMS / DOMjudge
- CNOI contestant self-testing packages, including automatically generated full-contest PDF statements and samples, as well as a CCF-style
self evaltesting program
Hull also provides automatic statement visualization based on Typst. Sample input/output, subtask tables, and Trait feature matrices can all be automatically extracted from judging data and injected into the statement. With built-in CPLib tracing and parsing, Hull can analyze a Validator's input-reading process and automatically draw graph/tree structures from sample cases, or apply Codeforces-style coloring for multiple test cases.
Blazing-fast CLI and automated toolchain: Hull includes high-speed stress testing powered by Rust + Rayon, using multi-core parallelism for concurrent data generation and validation. It can automatically capture Hack cases and generate formatted configurations. Local judging and compilation require no uploads to an OJ, allowing you to fully utilize high-core-count local machines for parallel compilation and execution, greatly accelerating iteration.
Built for the AI era: Hull is among the first to natively support Agent Skills and llms.txt. You can directly provide Hull's Skill index to DeepSeek / GPT / Claude and have an AI Agent automatically create problems while strictly following Hull's specifications!
- Official documentation and tutorials: https://hull.aberter0x3f.top/
- GitHub open-source repository: https://github.com/rindag-devs/hull
