I started this in October 2026 with my own children in mind. It is a road safety programme for children aged 6 to 11 and the parents who teach them, and I’m building it in the open as I go.
Each skill is a module. A module runs as a block of weekly sessions with four parts: a two-minute briefing for the parent, about ten minutes of rehearsal in the browser with an adult alongside, a walk at a real roadside, and a short debrief. The pedestrian skills come first, in the order Kerbcraft teaches them: choosing a safe place, then parked cars, then junctions. Cycling follows, working towards Bikeability Level 2. Two bonus modules sit alongside: using a push scooter, and being a car passenger.
The walk is the part that changes how children behave near a road. The game is there to rehearse it. It is never meant to replace the walk, and never meant to be used near a road.
Why build it
I looked for something to use first. I scored the UK options against a framework and set the bar at 15 out of 22. The best were Arility and Tales of the Road, at about 12 and 11. None covered both walking and cycling for this age group, and none rewarded stopping and looking over speed, so I decided to build.
Safety first
Parents will trust what this teaches and repeat it at the kerb, and a wrong rule could hurt a child on a real road. So safety is the main design problem, not a feature. Every rule traces to a source: the Highway Code, the National Standard for cycle training, the Kerbcraft manual, Know Your Traffic Signs or RoSPA guidance. There is a hazard log of thirty hazards, written in the shape of a clinical safety case. Build checks fail if a rule has no source, if the game shows a timer or a score, or if any copy says a child has passed or is ready. The game only corrects something a source prohibits.
Where it stands
The first module, “Safe place, stop, look, listen”, is built and plays end to end on an iPad and a PC. It is a static web app with no accounts and no tracking, and progress stays on the device. My family has tried it. The proper playtest with my own children comes next, and what it shows decides what gets built after that. Everything on these pages is as at late October 2026.
In these pages
- Safety by design: the rules the game never breaks, and what it will and won’t correct
- The hazard log: thirty hazards, how they’re scored, and four that changed the design
- Why this exists: the gap in UK products and what the evidence says about screens
- How a module works: the four parts, a session, and how a 6-year-old shows a reason without reading
- Decisions and dead ends: what changed along the way, and what is still open
- The build: the stack, content as data, and the checks that run on every change
- How the project runs: the repo, the Project, the build sessions and the credit budget