AI agents mark RLS 'done' with USING(true) policies that leave data wide open

Real_KingZeotic · reddit · 2026-09-23

A Reddit post highlights a common security blind spot in agent-built apps, especially on lovable/bolt/supabase stacks: verification only checks that a write landed, not that protection actually works.

Typical failure: an agent enables RLS on a table, writes, reads it back, sees it exists, and marks the task done—but the policy it generated is USING (true), so RLS is technically on yet every row is readable by anyone. The same happens with 'private' storage buckets that remain world-readable and functions running with elevated access and no internal auth check. The author argues most verification pipelines stop one level too early and should verify what a permission actually does after the write.

Original post →

More from coding & agent

coding & agent channel →