RetroSweep
Classic Minesweeper, made for a spare minute.
- Role
- Design and development
- Team
- Solo
- Timeline
- September 2026 – present
- Platform
- Web
- Status
- Live
- Problem
- A game of Minesweeper takes minutes and ends on one wrong click, which is a lot to ask of someone with a spare minute.
- My role
- I designed and built it solo: game UX, visual design, the token system and the front end.
- Key decision
- I added Safe Pick: five quick boards, each with exactly one safe tile to find, scored on speed and accuracy.
- Result
- Live at retrosweep.com, with Lighthouse scores of 100 for accessibility, best practices and SEO.
The problem
Minesweeper hasn't changed in decades
Minesweeper is one of the best-known puzzle games, and most versions still play the same way: one large board, several minutes of play, and a single wrong click ends the game. That suits a dedicated player, but not someone with a minute to spare. I wanted to keep what makes it satisfying, reading the numbers to find the safe tile, and offer a different way in.
How other versions compare
| Game | Quick mode | Daily boards | Streaks and leaderboard |
|---|---|---|---|
| Google Minesweeper | No | No | No |
| Microsoft Minesweeper | No | Yes | Yes |
| Minesweeper.online | No | Yes | Yes |
| RetroSweep | Yes | Yes | Yes |
Who it's for
One game, three ways to play
Minesweeper players don't all want the same thing, so each mode is built for a different one.
- Quick-break players want a full game in a minute or two. Safe Pick gives them five boards, one safe tile each, scored on speed and accuracy.
- Daily regulars want a reason to come back and something to compare. The daily gives everyone the same three no-guess boards, one scored attempt each, with streaks and a leaderboard.
- Traditional players want the game they know. Classic keeps the original rules, with a safe first dig, your time and your personal best.
Decision 01
A full game in five boards
Safe Pick shows five boards in a row, each with exactly one safe tile. Dig it and the next board appears. Your score is based on time and on good and bad digs.
Why
A classic game takes minutes and ends on one mistake. Safe Pick keeps the core skill, reading the numbers to find the safe tile, but fits it into a short round where a wrong dig costs points instead of the game.
Decision 02
Three no-guess daily boards, the same for everyone
Three no-guess boards a day, the same for every player, with one scored attempt each. Each board highlights a safe tile to start from.
Why
When everyone plays the same boards, a score means something: the leaderboard compares skill, not luck. The highlighted start tile, no-guess boards and a single attempt keep that fair. Nobody loses to a blind first click or a forced 50/50, and nobody can retry until they get it perfect.
Design system
Classic bevels, modern tokens
The game's core colours, fonts and sizes live in 40 named tokens, so changing one updates every screen that uses it.
How the tokens are organised
- 01Chrome and bevel
--face--light--shadow - 02Ink
--ink--ink-muted--ink-faint - 03Accent (navy)
--navy--navy-light--navy-tint - 04LED and status
--led-red--flag-red--streak-orange - 05Dark grounds
--bg--header-dark--text-dark
- Design tokens
- 40
- Components
- 26
Accessibility
Playable without a mouse, colour or motion
- Keyboard play: arrow keys move, Space digs, F flags.
- Touch: long-press to flag, or switch between 44px Dig and Flag buttons. Wide boards pan and pinch-zoom.
- Contrast: text tokens on the grey game surface measure 4.9:1 to 9.6:1, above the WCAG AA minimum of 4.5:1.
- Reduced motion: the chord, start-tile and daily pulses switch off. State never relies on colour alone: depth, fill and ARIA state all reinforce it.
- Lighthouse: 100 for accessibility, best practices and SEO on mobile and desktop (performance 85 mobile, 96 desktop), October 2026.
By the numbers
- Design tokens
- 40
- Components
- 26
- Lighthouse accessibility scoreMobile and desktop, October 2026
- 100
- Games played in the first month
- ~100
Lesson
Looking back
What I learned
I built RetroSweep in stages: Classic with a leaderboard first, then the daily, then Safe Pick. Building each mode was the easy part. The hard part was working out what would work as a product, and how to show three modes side by side. The SEO pages took more figuring out than I expected too.
What I'd do next
Test each new mode idea with a few players as a rough prototype before building it, and plan the SEO pages alongside the game from the start.