Comprehendo: A Generative AI-Powered Multilingual Reading Comprehension Training Platform

Comprehendo is an open-source multilingual reading comprehension training application. Its core feature is using Google Gemini AI to generate customized reading materials, supporting 16 languages and 6 CEFR levels. The platform offers features like instant feedback, text highlighting, word translation, and text-to-speech to help users improve their multilingual reading skills. Built with a modern tech stack and deployed on the Cloudflare Workers edge platform, the project has both practical utility and technical demonstration value.

• Original Author/Maintainer: tre-systems organization
• Source Platform: GitHub
• Release Date: June 12, 2026

Comprehendo aims to help users improve their multilingual reading skills. It uses the Google Gemini large language model to generate customized reading materials and practice questions, supporting a complete learning path from beginner to proficiency levels. The tech stack includes Next.js 15, React 19, TypeScript, Tailwind CSS, and it is deployed on the Cloudflare Workers edge computing platform.

Multilingual Support

Support for 16 languages (including Chinese, English, Japanese, Korean, etc.), with an internationalized (i18n) interface to provide a native-language operation experience.

CEFR Grading System

Strictly follows the 6-level CEFR standard (A1-C2), with each level corresponding to different language complexity to ensure materials are suitable for the user's level.

AI-Driven Content Generation

Uses Google Gemini 2.5 Flash to dynamically generate unique reading passages and multiple-choice questions, providing instant feedback, answer explanations, and text highlighting to enhance learning.

Auxiliary Learning Tools

• Word Translation: Hover the mouse to view (Google Translate API)
• Text-to-Speech: Listen to the pronunciation of articles/words (Web Speech API)

Frontend and Deployment

Adopts Next.js15 App Router + React19, uses TypeScript to ensure code quality, and Tailwind for responsive design; deployed on Cloudflare Workers (OpenNext framework + D1 SQLite) to reduce latency and support automatic scaling.

Authentication and Data

Supports GitHub/Google/Discord login via next-auth, and user data is stored in SQLite for cross-device synchronization.

Cost Management

• IP rate limiting (100 practice requests per hour)
• Caching strategy (AI-generated content stored in the quiz table)
• Daily AI request limit (1000 times) and independent limits for translation requests

Engineering Standards

ESLint + Prettier ensure code standards, Playwright for end-to-end testing, GitHub Actions for CI/CD; supports PWA installation.

Language Learners

Provides personalized training, AI generates fresh content to avoid memorizing answers, and CEFR grading allows for progressive learning.

Educational Institutions

Serves as a supplementary classroom tool: students practice independently, teachers understand progress via the management panel, reducing grading burdens.

Open Source Community

Provides modern web development examples (Next.js, Cloudflare Workers, AI integration, etc.) for developers to learn and practice.

Comprehendo combines generative AI with language learning needs, making it a practical open-source project with technical demonstration value. Its multilingual support, grading system, and engineering practices provide dual resources for language learners and developers.

In the future, as AI technology develops, similar applications are expected to play a greater role in personalized education; the open-source nature allows the community to contribute new languages, optimize AI prompts, and improve the interface, jointly promoting the project's evolution.