Build on Postgres.
Let your agents run it.

import { createClient } from "@supabase/supabase-js";

const supabase = createClient("https://shop.base.example.com", ANON_KEY);

const { data: orders } = await supabase
  .from("orders")
  .select("id, status, order_items(qty, products(name))")
  .eq("status", "paid")
  .order("created_at", { ascending: false });

// email, phone, OAuth, any OpenID Connect provider, anonymous users
await supabase.auth.signInWithOtp({ phone: "+15551234567" });
await supabase.auth.signInWithOAuth({ provider: "keycloak" });

// authenticator-app MFA; RLS can require (auth.jwt()->>'aal') = 'aal2'
const { data } = await supabase.auth.mfa.enroll({ factorType: "totp" });

supabase
  .channel("room-1")
  .on("postgres_changes", { event: "INSERT", schema: "public", table: "messages" },
      (payload) => render(payload.new))          // only rows RLS lets you see
  .on("presence", { event: "sync" }, () => showOnline())
  .subscribe();

await supabase.storage
  .from("avatars")
  .upload(`${user.id}/me.png`, file, { upsert: true });

const { data } = await supabase.storage
  .from("invoices")
  .createSignedUrl("2026/inv-0042.pdf", 3600);

-- turn on pgvector from the Extensions page, then:
create table documents (id bigserial primary key, body text, embedding vector(1536));
create index on documents using hnsw (embedding vector_cosine_ops);

select body from documents order by embedding <=> $1 limit 5;