Teachers can now take lightning-fast classroom attendance by capturing or uploading class group photos, letting an AI system automatically match faces to registered students. The update includes new modals for initial student face registration and a streamlined quick attendance flow. Additionally, past class sessions now skip the AI flow and default directly to manual editing to maintain accurate retroactive records. Overall, this introduces a huge quality-of-life win for teachers, drastically reducing time spent on daily roll calls.

We've cranked up the max output tokens for our Gemini face-matching endpoint from 1024 to 8192! 📸 This ensures you can process attendance for much larger group photos without worrying about getting truncated results. 🚀

On the main branch, subh05sus refactored the code structure for better readability and maintainability, updating API routes to include explicit assignments, homeworks, and courses fields in JSON responses. They also overhauled the OAuth consent page—adding scoped labels, icons, and a fresh design—and cleaned up the Prisma seed script by commenting out legacy boilerplate.

subh05sus pushed a major feature to main that implements full OAuth2 integration with Sparkmentis—adding /oauth/authorize and /oauth/token endpoints with PKCE, JWT access and hashed-refresh tokens, plus read-only integration APIs for user, courses, assignments, and homeworks. They also built an outbox/webhook system (with a cron dispatcher), added scope validation and signature verification tests, updated .env.example and seed scripts, and shipped docs for the Sparkmentor integration.

subh05sus pushed a new health‐check endpoint for the Google Gemini AI integration on the main branch. The added GET route in app/api/health/ai/route.ts probes the Gemini 2.5-Flash model via @google/genai, handles missing API keys and errors gracefully, and returns a simple JSON status with 200 or 503 codes. This will help you monitor service availability and configuration in a lightweight way.

subh05sus pushed to main, integrating Sentry for error tracking and monitoring by updating configuration files and adding the Sentry dependency. They also bumped @sentry/nextjs from v8.46.0 to v8.55.0 in package.json.

On the main branch, subh05sus updated the AI integration code by replacing all instances of the “gemini-2.5-flash-lite” model with “gemini-2.5-flash” and standardized string quoting across numerous files. This ensures the project now consistently uses the newer Gemini AI model throughout.

subh05sus merged the sparkmentis:main branch into main and cleaned up the ParentMessagesPage by removing an unnecessary null fallback for selectedConversationId. They also refactored and standardized code formatting across admin/teacher parent-verification routes and UI components, enhancing error handling throughout.

subh05sus merged updates into main, refactored UI components for better responsiveness and introduced a new parent dashboard. Rohit-Dnath added Terraform-based GCP infrastructure for Cloud Run and Cloud SQL and streamlined the GitHub Copilot instructions for clarity.

On the main branch, subh05sus merged in sparkmentis’s updates and Rohit-Dnath added a full “Assignment Grader” prompt template (with rubric support, dynamic fields, UI enhancements and backend wiring in AIGradingService), along with some basic UX improvements like custom scrollbars, prompt-editor tweaks and a route rename.

subh05sus merged sparkmentis:main into the main branch, bringing in Rohit-Dnath’s update to the semester dropdown. The change bumps the select options from Fall 2024–Winter 2025 up to Fall–Winter 2026.

subh05sus merged in the latest updates from the sparkmentis main branch, and Rohit-Dnath added new dropdown options on the course creation page—introducing “Science” as a subject and “Winter 2025” as a semester choice.

subh05sus pushed two merge commits on main—one bringing in Rohit-Dnath’s changes from the sparkmentis repo—and Rohit-Dnath’s latest update populates the course-creation dropdown with full K–12 grade options.

subh05sus merged sparkmentis:main into main and refactored several API routes (student late‐request, teacher late‐requests, and grading) to handle route parameters as Promises. Now each handler awaits params (e.g. params.id, params.submissionId) for more reliable ID extraction.

subh05sus pushed to the main branch and tweaked the README header formatting for a cleaner display of the SparkMentis project title.