An interactive reading comprehension game using AI to generate cloze (fill-in-the-blank) exercises from public domain literature. Deployed on Cloudflare Workers via Kale Deploy.

Live: cloze-reader.cuny.qzz.io

Passages from Project Gutenberg are streamed via the Hugging Face Datasets API. Gemma-3-27B (via OpenRouter) selects contextually meaningful words to remove, producing fill-in-the-blank exercises that test reading comprehension through contextual inference. Players can consult structured hints about part of speech, sentence role, and synonymy without being given the answer directly.

• Runtime: Cloudflare Workers (Hono, TypeScript)
• Frontend: Vanilla JS, ES6 modules, Tailwind CDN — no build process
• AI: Gemma-3-27B via OpenRouter, proxied server-side at /api/ai/chat
• Data: Project Gutenberg via HF Datasets API, cached in Workers KV
• Storage: D1 (SQLite) for leaderboard and gameplay analytics

npm install
npm run dev          # wrangler dev — local Workers runtime
npm run check        # TypeScript type check

Create .dev.vars for local secrets:

OPENROUTER_API_KEY=sk-or-...
HF_TOKEN=hf_...

Push to main triggers Kale Deploy → cloze-reader.cuny.qzz.io. Secrets are set via the Kale MCP.

Ported from milwrite/cloze-reader (Python FastAPI + Redis on Railway) to Cloudflare Workers with D1/KV.