Better Bet
A multi-agent betting picks management system
- Bun
- Hono
- SQLite
- React
- TanStack Router
- Tailwind
- shadcn/ui
Sports betting picks management with an agent API, real-time SSE activity feed, and analytics dashboard.
Problem
How do you build a system where multiple AI agents submit structured predictions, and you track performance across them with real-time visibility?
Approach
The backend runs Bun with Hono and Bun’s native SQLite. Admin routes use session-based auth (HttpOnly cookies), while agent routes authenticate via per-agent Bearer tokens with rate limiting at 100 requests per minute.
Agents submit picks through documented batch endpoints - single pick creation, batch submission with 207 Multi-Status responses, batch closing line updates, and batch result settlement. The system auto-migrates on startup in WAL mode.
A real-time SSE activity feed at GET /api/activity/stream pushes updates as agents submit picks, settle results, or update closing lines. The frontend (Vite + React + TanStack Router + shadcn/ui) provides an analytics dashboard with per-agent, per-market, and per-competition breakdowns plus daily P&L tracking.
Outcome
A working multi-agent system where agents authenticate, submit, update, and settle picks through a documented API - not a demo, but an operational picks tracker.