Is my Supabase database public?

Probably not, but the only way to know is to look from outside, the way a stranger would. It takes one request and no tools.

Why this question comes up at all

Supabase serves every table in the public schema over a REST endpoint, and your anon key ships inside your frontend bundle on purpose. Anyone can read that key out of your JavaScript in a few seconds. That is not the flaw.

The flaw appears when a table has no Row Level Security. Then the anon key, which everyone has, is enough to read and often write that table. The key was never the secret. The policies are the protection.

Check it in 30 seconds

Open your deployed site, open developer tools, and search the loaded JavaScript for supabase.co. You will find your project URL and anon key. Then, from a terminal, ask the database directly:

curl "https://YOUR-PROJECT.supabase.co/rest/v1/YOUR_TABLE?select=*&limit=5" \
  -H "apikey: YOUR_ANON_KEY"

Read the answer carefully:

Do the same for writing, which people forget to check:

curl -X POST "https://YOUR-PROJECT.supabase.co/rest/v1/YOUR_TABLE" \
  -H "apikey: YOUR_ANON_KEY" -H "Content-Type: application/json" \
  -d '{"id":"00000000-0000-0000-0000-000000000000"}'

Anything other than a refusal means a stranger can insert rows into your database.

What the dashboard does and does not tell you

The shield beside a table tells you Row Level Security is enabled. It does not tell you whether the policy restricts anyone. These two tables both show a shield, and one of them is wide open:

-- Protected: a row belongs to the user who owns it
create policy "own rows" on public.invoices
  for all to authenticated
  using ( (select auth.uid()) = user_id )
  with check ( (select auth.uid()) = user_id );

-- Shield on, protection off: this passes for every caller, signed in or not
create policy "enable read" on public.invoices
  for all using ( true );

The second policy is common because it is the fastest way to make an error message go away. It is worth searching your migrations for using (true) before you trust any shield.

The other half: privileges

Row Level Security is checked after table privileges, not instead of them. A grant like this hands write access to every anonymous visitor regardless of your policies:

grant all on public.invoices to anon;   -- almost never what you want

List what the anonymous role can actually do:

select table_name, privilege_type
from information_schema.role_table_grants
where grantee = 'anon' and table_schema = 'public'
order by table_name;

If you find something open

Turn Row Level Security on before you write the policy. An app that errors is better than a table that leaks, and the error tells you exactly which query needs a policy:

alter table public.invoices enable row level security;
alter table public.invoices force row level security;
revoke all on public.invoices from anon;
grant select, insert, update, delete on public.invoices to authenticated;

create policy "own invoices" on public.invoices
  for all to authenticated
  using ( (select auth.uid()) = user_id )
  with check ( (select auth.uid()) = user_id );

Then check from outside again with the same curl command. Trusting the dashboard instead of the response is how people stay exposed after fixing it.

If a service key was also exposed, rotate it before anything else. Deleting the line does not help, because the old value stays in your git history.

Related


Check your own project

One command. No install, no signup, and it never touches the network. It reads your repository and tells you which tables are reachable and which keys are exposed, with the fix beside each finding.

npx github:humora2504/vibeproof .

Free and MIT licensed. If you want the policies generated for your own schema, and a script that proves an anonymous visitor is refused, there is a $19 Fix Pack. The scanner stays free either way.