A sophisticated AI-driven Mafia game simulation where LLM agents play against each other.

• Monorepo: Managed with pnpm workspaces.
• Frontend: React (Vite) in frontend/.
• Backend: Cloudflare Workers + Hono + Durable Objects in backend/.
• Database: Cloudflare D1 (SQLite).
• AI: Google Gemini (via Google Gen AI SDK).

• Node.js v20+
• pnpm (npm install -g pnpm)
• Cloudflare Wrangler CLI (npm install -g wrangler)
• GitHub CLI (brew install gh) - Optional, for repo setup.

Just run:

pnpm dev

This magic command will:

1. Check for secrets: It will try to fetch secrets or create a dummy .env if missing.
2. Sync Environment: It ensures the backend has the correct .dev.vars.
3. Start Everything: It launches both the Frontend (localhost:3000) and Backend (localhost:8787) in parallel.

To easily create a private GitHub repository and sync your local secrets:

./scripts/setup-github.sh

Run End-to-End Tests:

pnpm test:e2e

Pre-commit Hooks: Husky is configured to run linting on staged files before commit.

Deployment is automated via GitHub Actions on push to main.

To deploy manually:

pnpm deploy