Works with the AI Agents you already use
Supabase RLS Launch Gate (Pre-Launch Security Check)
Anyone holding the public anon key can then read or write every row.
$9
About this skill
The problem\n\nThe most common Supabase breach is simple: a table in public with Row Level Security off, or a policy written as using (true). Anyone holding the public anon key can then read or write every row. It is easy to miss across dozens of migrations.\n\n## What it does\n\nYour agent replays your supabase/migrations in order (so a later DROP POLICY or DISABLE ROW LEVEL SECURITY counts) and returns a PASS / BLOCK verdict with a per-table status and file:line evidence for each finding.\n\nBlockers\n- public table with RLS off\n- Write policy (insert/update/delete) that is true for anon or authenticated\n- Policy that trusts user_metadata, which the signed-in user can edit\n\nWarnings\n- Public read of all rows (fine if intended, but check for private columns)\n- Policy that only checks auth.role() = 'authenticated' (every signed-up user can act on every row)\n- UPDATE WITH CHECK that does not tie the row to its owner\n- Views without security_invoker = true (they skip RLS)\n- SECURITY DEFINER functions without a pinned search_path, and ones exposed in public\n\nEvery finding includes the fix, written as a migration for you to review, such as owner-scoped policies using (select auth.uid()) = owner.\n\n## Safe by design\n\nIt only reads .sql files. It never connects to your database, never runs SQL and never applies fixes. Works on migrations or on a schema dump you make yourself with supabase db dump.\n\n## Checked against a real database\n\nThe bundled test fixtures were applied to a real PostgreSQL 16 server with Supabase-style anon and authenticated roles, and then actual reads and writes were attempted as each role. All 8 verdicts matched what the database really allowed. 17 further tests cover parsing edge cases (dollar-quoted function bodies, quoted identifiers from db dump, replaced policies, views and functions).\n\n## Requirements\n\nPython 3.8+ standard library only. No network, no packages.\n\n## Limits\n\nStatic analysis of SQL text: dashboard edits not in your migrations are invisible (use a dump). Column-level grants, Storage (storage.objects) policies, Edge Functions and service-role key usage are not analysed. Policy conditions are classified by pattern, not evaluated, so a PASS is not a full security audit.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 month ago
Requires