Aster

A multilingual voice + chat assistant that helps students research, verify, and fill government application forms (any scholarship; verified rule packs for MahaDBT, NSP and LIC).

Start here: docs/SETUP_GUIDE.md. Then open Claude Code in this folder and run /status, then /milestone M0.

Path	What
CLAUDE.md	Context Claude Code loads every session (stack, rules, guardrails)
docs/	Product, architecture, agent, voice, research, verification, form-fill, frontend, API specs + build plan + setup guide
.claude/commands/	/milestone, /guardrails, /status
supabase/migrations/0001_init.sql	Full schema with RLS, storage bucket, audit hash chain, endpoint registration
gpu/aster_gpu_worker.ipynb	Kaggle 2×T4 worker (LLM/vision + ASR + OCR behind one authenticated gateway + Cloudflare tunnel)
render.yaml	Reference copy of the Render backend service settings (live service was created manually; web goes to Vercel)
ops/phone/	Optional backup: Termux scripts to run the backend + tunnel on the phone
knowledge/	Verified scheme packs (template inside)
backend/, web/	Built by Claude Code from M0 (env examples included)

Run locally

# backend → http://localhost:8000/health
cd backend && cp .env.example .env && uv sync
uv run uvicorn app.main:app --reload --port 8000
uv run pytest -q && uv run ruff check . && uv run ruff format --check .

# web → http://localhost:3000
cd web && cp .env.local.example .env.local && pnpm install
pnpm dev
pnpm lint && pnpm typecheck && pnpm build

Deploy: push to main → Render redeploys the backend (https://aster-jj5b.onrender.com); setup in docs/SETUP_GUIDE.md Phase 5b.