Your scoring system,
documented for you.
When the agent inspects your repo it doesn't just write config — it produces this: a living document of every scored action, the journeys they form, and how the live economy tracks against your targets. Here's an example for a focus & habit app.
Features & user actions
Three product areas were detected. Each user action maps to a scored event, with anti-gaming caps the agent inferred from how often the action can realistically happen.
Focus sessions
src/features/focus/Reflections
src/features/journal/Routines
src/features/routines/A mapped journey
The agent stitches actions into the journeys your users actually take, and shows the score accumulating through the loop — including bonuses.
How the economy behaves
Score progression for a typical user, and where points actually come from.
Cumulative score · first 30 days
Points by source
Metrics vs target
The agent tracks the outcomes your economy is meant to drive, against the targets you set — and flags where a scoring change might help. This is what closes the loop on tuning.
Suggested change. Day-7 streak survival trails target. The agent proposes issuing the first freeze token at a 5-day streak instead of 7 — drafted as scoring v5 for your review.
Generate this for your app.
Point the agent at your repo and it produces this document — then keeps it live as your product and your numbers change.