ESP32-S3 Bench Guide
ESP32-S3 N16R8 bench guide: live pinout, lessons, simulators and the bench log.
翻译准备中
尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证正在加载完整 README。 以下为已保存的摘录。
Kalinga is an offline-first teacher assistant prototype for multigrade classrooms. It combines class and learner management, date-specific attendance, ILAW lesson planning, shared resources, and Gabay—a page-aware Taglish teaching guide.
Requirements: Node.js 22.13 or newer.
Open http://localhost:3000. The interface and local Gabay guidance work even before the backend values are added.
Create a Supabase project, then copy its Project URL and publishable key into .env.local:
Never place the Supabase service-role key in the browser environment or commit it to Git.查看完整 README ↗Kalinga is an offline-first teacher assistant prototype for multigrade classrooms. It combines class and learner management, date-specific attendance, ILAW lesson planning, shared resources, and Gabay—a page-aware Taglish teaching guide.
The production app prepares its application files for offline reopening. Regular teachers can reopen their saved workspace on the same personal device, use saved lessons and attendance, and explicitly download PDFs for offline reading. Gabay AI and community features still need internet. Installing the PWA is optional; normal development does not register a service worker.
Read offline preparation, sign-out and verification limits before testing. Production-browser tests use synthetic Supabase fixtures, not real accounts. After building, run npm run test:offline with installed Chrome (or PLAYWRIGHT_CHROMIUM_EXECUTABLE_PATH). Actual Android and live-account acceptance remain separate checks.
Requirements: Node.js 22.13 or newer.
npm install
cp .env.example .env.local
npm run dev
Open http://localhost:3000. The interface and local Gabay guidance work even before the backend values are added.
Create a Supabase project, then copy its Project URL and publishable key into .env.local:
NEXT_PUBLIC_SUPABASE_URL=https://your-project.supabase.co
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY=your_publishable_key
Never place the Supabase service-role key in the browser environment or commit it to Git.
Link this repository to the project and apply the schema:
npx supabase login
npx supabase link --project-ref YOUR_PROJECT_REF
npx supabase db push --dry-run
npx supabase db push
The migration creates teacher-owned profiles, classes, learners, lesson plans, attendance, resources, a private resource bucket, and row-level security policies. Each authenticated teacher can access only their own classroom records; only explicitly shared resource metadata is readable by other teachers.
Gabay calls Groq only through a Supabase Edge Function, so the Groq key is never shipped to the browser. Add the secret and the allowed web origins:
npx supabase secrets set GROQ_API_KEY=YOUR_GROQ_KEY
npx supabase secrets set GROQ_MODEL=openai/gpt-oss-20b
npx supabase secrets set ALLOWED_ORIGINS=https://your-vercel-domain.vercel.app
npx supabase functions deploy gabay-chat
Only the origins in ALLOWED_ORIGINS may call the function. To develop against a
deployed function, opt localhost in explicitly — leave this unset in production:
npx supabase secrets set ALLOW_LOCAL_ORIGINS=true
Each teacher may make 30 Gabay requests every 5 minutes. The counter lives in
public.gabay_rate_limits and is only writable through the claim_gabay_request
function, so a browser cannot reset its own quota. The check fails open: if it
cannot run, the request is logged and allowed, because the limiter guards cost
rather than access and should never be able to take Gabay down. Override the
defaults with:
npx supabase secrets set GABAY_RATE_WINDOW_SECONDS=300
npx supabase secrets set GABAY_RATE_MAX_REQUESTS=30
Connected Gabay requires a valid Supabase user session. Prototype mode does not call the AI service or expose the secret.
The AI receives only the recent chat and minimal classroom context: the teacher’s preferred display name, current screen, grade levels, subject, lesson topic, and connection state. Saved learner names, reference numbers, attendance statuses, and notes are deliberately excluded.
Add these two values to the Vercel project’s Environment Variables and redeploy:
NEXT_PUBLIC_SUPABASE_URLNEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEYKeep GROQ_API_KEY in Supabase Edge Function secrets, not Vercel’s public environment and never in source control.
npm run lint
npm test
npm run build
Migrations can be checked without Docker or a Supabase project. This applies every migration to a throwaway local Postgres and asserts the behaviour they produce — mention tags, notification triggers, the backfill, Gabay rate limiting, and that row-level security still isolates one teacher from another:
brew install postgresql@17
./scripts/verify-migrations.sh
It never touches a real database. Run it before supabase db push.