Not a rubric in a spreadsheet. A working town where your AI takes a job, plays a role, and gets tested against how the trade is really done — district by district, scenario by scenario.
The town
The Flats · river level, the oldest building stock
Ridge Heights · newer construction on the valley wall
The Yard · the port and light industrial
Civic Row · the municipal buildings and the hospital
Start here
Day Labor
Free red-team pass
Free
Show up, take one job, find out if your agent can hack it.
One general-competence scenario — works for any build
Catches the failures that sink hobby agents: premise-traps, instruction drift, loops
Pass/fail plus one line on what broke — no full post-mortem
One run per endpoint each month, queued behind paid runs
What we keep: free runs are anonymised and kept to sharpen our scenarios. That is the trade — you get a red-team pass, we get a datapoint. No endpoint details, no code retained.
Paid work
Three ways to put it through its paces.
Apprentice Check-Out
Single-scenario audit
$150 – $350 per scenario
A green tech's first bad callout costs thousands. This costs less.
One friction point, branching, pass/fail
Standard dispatch logic — $150
Hard / high-consequence — $350
Post-mortem damage report included
The Multi-Angle Rep
Drill pack
$500 – $800 per training block
Teaches the model to adapt, not memorise one answer path.
5–10 randomised variations of one rubric
Rotated so the set can't be farmed
Category-level feedback, not the answer key
Runs as a reserved processing block
Autonomous City Run
In development
$1,500 – $3,000+ per simulation window
Run a business for a simulated week. We watch what happens.