CogniLoop is an intelligent teaching assistant system designed to help educators create and grade assignments using large language models. Teachers can upload course documents, generate question sets based on the content, and automatically grade student responses, providing scores and detailed explanations. This system aims to provide advanced AI educational tools to students regardless of location, acting as a supplementary learning resource.

How It Works

The system leverages large language models (LLMs) and Retrieval-Augmented Generation (RAG) to process course documents and generate relevant questions. Student answers are then evaluated by the LLM, which provides automated grading, scoring, and feedback. The technical stack includes FastAPI for the backend, PostgreSQL with the pgvector extension for data storage and vector embeddings, LangChain and LangGraph for orchestrating LLM workflows, and React with Vite and TypeScript for the frontend. Deployment is managed via Docker Compose.

Quick Start & Requirements

• Primary install/run command: Use Docker Compose. After cloning the repository (git clone https://github.com/itsmorninghao/CogniLoop.git) and navigating into it, copy .env.example to .env and configure necessary variables (like JWT secret). Then run cd docker-cogniloop && docker-compose up -d --build.
• Non-default prerequisites: Docker and Docker Compose must be installed. Configuration of LLM and Embedding models (API Key, Base URL, model name) is required via the admin backend.
• Setup: Initial setup involves cloning, configuring environment variables, building and starting services via Docker Compose, creating a super administrator account upon first access, and configuring LLM/RAG parameters.
• Links: Online Demo: https://cogniloop.morninghao.online. GitHub: https://github.com/itsmorninghao/CogniLoop.git.

Highlighted Details

• Automated generation of diverse question types from course documentation.
• AI-powered grading of student answers with immediate feedback and scoring.
• A "Question Square" feature allows teachers to share generated tests, with leaderboards for student participation.
• A planned "Gaokao" (National College Entrance Examination) question generation feature requires specific data import and configuration.

Maintenance & Community

The project is actively under development, with plans to continuously improve features and user experience. Users are encouraged to submit issues via GitHub for support and bug reporting. No specific community channels (like Discord or Slack) are listed.

Licensing & Compatibility

The project is licensed under AGPL v3. This strong copyleft license requires any derivative works or modifications distributed to users to also be made available under the AGPL v3 license. This may impose restrictions on integration with closed-source or proprietary systems.

Limitations & Caveats

CogniLoop is currently in development, and some advanced features, such as personalized student learning paths based on performance analysis, are still planned. The "Gaokao" question generation functionality requires additional setup steps beyond the basic deployment. The AGPL v3 license necessitates careful consideration for commercial use or integration into existing proprietary software stacks.