TsundereSQL by Orphan Rows

itch.io

7 min read Original article ↗

Wishlist TsundereSQL on Steam. Full game Nov 12.

thank you to everyone who supported tsunderesql here on itch.

the full course, all 14 modules, is free here until the steam launch on nov 12.

download the windows build before then to keep the full course and every lesson. after nov 12 the windows download is removed and the browser version becomes modules 1 to 3.

progress in the browser stays in that browser.

keys: everyone who paid any amount before oct 19 gets a free steam key for the full game. from now until nov 12, payments of $10 or more also get one.

keys are emailed around launch to the email you shared with itch at checkout. didn't share one? email hello@tsunderesql.com.

Solve real SQL. The database runs your query, checks the answer against hidden data, and tells you if you got it wrong. Get it right and the story moves forward.

TsundereSQL is a browser-playable anime visual novel built around a full SQL course. You study under a trio of instructors who set the problems, grade your work, and slowly warm up as you improve. Every SQL exercise runs locally in your browser on DuckDB. There is no account and no install, and SQL execution and grading stay in the browser. You type SQL, real rows come back, and a hidden test decides if the answer holds.

It runs in the browser and is free here until Nov 12, all 14 modules. After that, Modules 1 to 3 stay free here and the full course moves to Steam. Open it and start Problem 1.

▋ WHAT YOU ACTUALLY LEARN

224 authored problems across 14 modules, all free here until Nov 12 (then Modules 1 to 3 here, all 14 on Steam). 220 are SQL exercises that run and grade locally against seeded DuckDB datasets, each with 2 to 4 hidden test cases, not multiple choice. Four are open-ended analyst judgment cases:

  • SQL Foundations and Aggregation & GROUP BY to get you querying and grouping without the "column must appear in the GROUP BY clause" wall.
  • Joins & Set Operations: the topic the whole SQL community fights about. Left, right, inner, anti, the lot, on real tables.
  • Subqueries & CTEs for breaking a hard question into readable steps.
  • Window Functions, split across two modules: ranking and QUALIFY, then frames, gaps and islands, and sessionization.
  • Product & Growth Analytics: retention, cohorts, funnels, the queries an analyst actually ships.
  • Advanced SQL (recursion, JSON, reshaping), Working with Data at Scale, Data Modeling,

Working With Production Data, and Thinking Like a Senior Analyst for the messy real-world parts: dates stored as strings, NULL that is not zero and not empty, numbers that were wrong before you arrived.

  • The Growth Desk and The Interview Gauntlet: an interview-style gauntlet of questions to prove you can do it under interview pressure, not just with the docs open.

If you already write SQL, start at the module that hurts. If you are starting from zero, start at the top and the difficulty climbs with you.

▋ YOUR INSTRUCTORS

Three instructors run the course, and the game is as much about them as it is about the SQL. They have their own standards, their own tempers, and their own reasons for being hard on you. Aimi joins after you clear Module 1, Nora after Module 2, so each finished module hands you a new teacher.


Tsumugi - senior analyst, reluctant tutor. She sets the problems, marks them hard, and would like it on record that she is not doing this because she enjoys it.

Aimi - the warm one. Endlessly patient, genuinely proud of you, and the only person here who will tell you a LEFT JOIN is kind.

Nora - cold, precise, devastating. "It works. It's also O(n²). Redo it." Impressing her takes a while. It is worth the while.

▋ AND THE PEOPLE AROUND YOU

The academy is not just the faculty. Six of the people you meet have their own routes: their own scenes, their own arc, their own ending. You get there by doing the work and spending your time with them, the same way everything else here is earned.


Mei - a peer, one step ahead, and she will not let you forget it. Teases like it is cardio.

Reina - heiress to the exchange, learning data so she can run it. "Make the board care, in one slide. Begin."

Mika - identity resolution. "Give me three events and I'll tell you everything." She has already run it on you. She would like you to think that is sweet.

Yume - cannot write a JOIN to save her life. Knows the dashboard is lying before it finishes loading. Nobody can explain how.

Vega - crashed into the quad to study the dominant data species. Takes you EXACTLY at your word, which is a bigger problem than it sounds.

Lorelei - the Siren of the Data Lake. Her song is the beautiful wrong answer, and it is always the query you most want to be right.

The writing plays the tsundere register straight. The cast is dry, exacting, and slow to admit they are impressed. When one of them finally softens, you earned it.

▋ WHAT THIS IS / WHAT THIS ISN'T

What it is: a real SQL course wrapped in a story, for people who learn better when something is at stake. The curriculum is complete and sequenced. The characters and romance routes are the reason to keep going. You choose who you spend your time with, and each instructor has a fixed, developer-directed route with its own scenes and ending.

What it isn't: it is not a branching-choice simulator where every decision reroutes the plot. The routes are fixed and developer-directed, and the story lands the way it is meant to land. It is not a reference manual or a cheat sheet either. You write the queries yourself, and the game checks them. And it is not job-training marketing: it teaches SQL well, but it is a game first.

▋ AI DISCLOSURE

AI disclosure: Character art, CGs, and backgrounds are AI-generated and human-curated. Your SQL runs on a real database in your browser and is graded against a reference query. The optional tutor and four free-text judgment exercises can call an LLM for feedback. All other story content is fixed.

▋ CONTENT NOTES

Content notes: swimwear, lingerie and corset styling, bunny-suit cosplay, and onsen/towel imagery; romance themes; heavy story themes (grief, memory loss). No explicit content. The suggestive reward art is a single toggle in Settings, and turning it off removes it from the gacha and the wardrobe.

▋ CREDITS

Designed, built, and directed by Orphan Rows. Character art and backgrounds are AI-generated and human-curated (see the disclosure above).

Music: 14 public-domain and Creative Commons piano and orchestral recordings, plus three character themes. Full track list, performer credits, and licenses are in the game under Settings > About, and in docs/AUDIO_PACK.md. Sound effects: Kenney (CC0).

Built with React, TypeScript, and DuckDB-Wasm. Every SQL exercise is graded by a real database running in your tab.

▋ FIND OUT MORE