This update streamlines how test case inputs and outputs are structured by moving from dynamic objects to string-based JSON representations. This change improves compatibility with Gemini's structured-output mode and introduces robust validation logic to ensure data integrity during test case editing and AI-driven regeneration. These improvements provide a more reliable and consistent experience when managing unit tests.

Updated test case definitions in lib/code-ai.ts to use string-based JSON representations instead of dynamic-key objects. This change significantly improves compatibility with Gemini's structured-output mode and introduces robust validation logic to ensure data integrity. These modifications should reduce parser errors in AI-driven test execution.

This update introduces automatic starter code generation for multi-language support in the problem-solving interface, alongside a major UI overhaul switching to a cleaner light theme. We've also improved Docker deployment by including necessary Prisma binary files and added a feature-gated sidebar navigation based on course availability.

This update introduces comprehensive management features for exams and unit tests, including new API endpoints and UI improvements for course administration. Additionally, we have integrated grade level support across student onboarding and course browsing for better academic alignment, along with bulk user promotion tools for administrators.

We have updated the Dockerfile to use Node.js 20-alpine across all build stages to leverage the latest performance and security enhancements. Additionally, the build script was simplified by removing the experimental turbopack flag to streamline our production builds.

This update introduces significant improvements to course and user management. Admin dashboard now supports semester filtering and role-based user searches, while teachers benefit from a new grouped view for courses. Additionally, student course pages have been refactored for improved performance and layout consistency.

Introduced a new management interface allowing admins to assign and remove multiple teachers per course. Key updates include a new API endpoint for handling teacher assignments and a 'ManageCourseTeachersDialog' component featuring role-based management and primary teacher protection. This significantly enhances course oversight capabilities for administrative staff.

We've enhanced the platform's navigation by adding Course Packs, Pack Enrollments, and Config items to the admin sidebar, along with Pack Enrollments for teachers. Additionally, this update resolves an issue where the admin course-pack detail page failed to properly parse the API response for course data. These changes streamline workflow management for course administrators and instructors.

Introduced a cron-based webhook handler that manages event queues with automatic retry logic and backoff. The system now tracks delivery status, attempt counts, and response details in the new WebhookDelivery and WebhookEvent schema models, ensuring reliable synchronization for integrations.

We have introduced a new admin dashboard designed to provide full visibility and control over webhook operations. This update adds specific endpoints for managing subscriptions, tracking event history, and reviewing delivery logs and metrics, streamlining the debugging and monitoring process for our webhook system.

We have transitioned to an asynchronous webhook delivery model to improve system reliability and responsiveness. The new implementation supports event queuing, automated retry logic for failed deliveries, and dedicated API endpoints for testing webhook subscriptions. This ensures better visibility into webhook health and increases robust delivery for external integrations.

Introduced a robust authentication layer for the V1 API that supports both API keys and OAuth tokens. The handler includes secure practices like constant-time hashing for API key validation, integrated IP allowlisting, and native rate limiting to ensure API stability and security.

This massive update introduces full AWS infrastructure as code using Terraform, enabling production-ready deployments on EC2 with RDS and Redis. Alongside the infrastructure, a complete API management administrative dashboard was built, featuring key rotation, OAuth client management, rate-limiting, and automated OpenAPI documentation generation. Finally, a robust webhook event emission system and CI/CD pipelines via GitHub Actions were implemented to streamline maintenance and integration processes.

This update introduces a dynamic sidebar that automatically adapts its branding based on the tenant's primary color, enhancing the multi-tenant UI experience. I also refactored the middleware to centralize tenant suspension checks and permission logic, ensuring a more resilient and performant authorization flow for all user roles.

We've added multi-tenant capabilities, allowing the application to resolve configuration dynamically based on the request domain. The update introduces a new TenantProvider for dependency injection and standardizes tenant-specific settings like branding and feature flags. This shift enables scalable, isolated environment configuration without infrastructure overhead.

Introduced a comprehensive UI for teachers to view student assignment submissions, track grading progress, and apply evaluations. The new view includes dedicated components for reviewing student work, managing assignment-level grades, and an AI-powered grading assistant with confidence scores and suggestion workflows to streamline feedback processes.

This update introduces a comprehensive multi-tenant architecture supported by a new feature-flagging system, allowing for better isolation and conditional capability delivery. Additionally, I've significantly expanded the AI-powered course management APIs to better support automated grading and planning workflows across the platform.

We've introduced a comprehensive multi-tenant infrastructure to the platform. This update includes middleware-based tenant slug detection and dynamic sidebar configurations tailored to specific tenants, providing better isolation and customization. This modular approach significantly improves scalability for multi-tenant deployments.

We've significantly upgraded the unit test experience for both students and teachers! Students now receive immediate AI grading and personalized feedback right after test submission, featuring beautifully redesigned overall scores and markdown-rendered feedback. For teachers, the grading workflow is now much smoother with a new "AI Grade All" button, instant generation of comprehensive feedback reports, and improved responsive layouts across the board.

We've extracted the automated MCQ and AI-assisted unit test grading logic into a dedicated module. This cleans up multiple API endpoints by replacing duplicated code, ensuring that grading rules stay consistent whether grading a single submission or in bulk. Moving forward, updates to our evaluation algorithms will only need to happen in one place!