A table in Supabase's public schema without row level security can be read and changed by anyone who has your anon key, and that key ships inside your app. This skill finds that and eight other Supabase SQL mistakes, and tells you how to fix each one.

Your agent runs a scanner on your supabase/migrations and schema files, in order, the way migrations run. RLS turned on in a later migration counts, renamed tables are followed, and supabase db pull output is understood. It reads policies (for, to, using, with check), views, security definer functions and extensions. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Keys are never printed.

What it checks

BLOCK (fix before shipping)

• R01: a public table without row level security

• R02: an insert, update or delete policy open to anon or public with true

• R03: a public view without security_invoker, which bypasses RLS

• R04: a policy that trusts user_metadata, which users can edit themselves

• R05: a service_role key or sb_secret_ key written in SQL

REVIEW (check before release)

• R11: a select policy open to everyone

• R12: a write policy where any signed-in user can change any row

• R13: a security definer function callable through the API

• R14: an extension installed in the public schema

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• Risky and safe example migrations, and the real report the scanner produces on the risky ones

Limits

It reads SQL files, not your live database, so changes made in the dashboard are not seen. Run the Supabase Security Advisor too. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 8 files: SKILL.md, the scanner in scripts/ (2 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (3 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.