Supabase Puts Every Service's Logs in One Searchable View

Supabase opened a beta on July 16 for Unified Logs, a single searchable view that pulls together logs from every service behind your project: the API gateway, Postgres, Auth, Storage, PostgREST, Realtime, and the connection poolers.

Why it matters

If you build on Supabase, especially through an AI coding agent or app builder that wires up your backend for you, debugging a broken request usually means guessing which service failed first. Was it the gateway rejecting the request? A Postgres query timing out? An Auth check that silently failed? Until now, tracking that down meant checking separate logs for each service, one at a time.

Unified Logs puts them in one place. You can filter by log type, level, status, method, or path, or search free text across everything. Filters live in the URL, so you can copy a link to exactly the filtered view you’re looking at and hand it to a teammate, or paste it into a support request. A severity-color-coded histogram shows request volume over time, and clicking into a single request opens a detail panel that traces it end to end: network timing, the gateway, the database query, and the Auth check, together.

There’s also a live tail. Clicking Live streams new events into the table as they arrive, useful for watching what happens in real time while you reproduce a bug.

Who should care

Anyone running a Supabase-backed app who has ever had to guess which service caused a 500 error. This is squarely a debugging tool, not a new API or SDK. It doesn’t change how you write code, but it changes how fast you can find out why something broke.

What builders should do next

Open the Logs section of your Supabase project dashboard. Supabase says the feature is rolling out to all projects during the open beta, and the classic Logs Explorer is still available behind a toggle if you want to compare. Try filtering by status code the next time a request fails, then click into that request’s detail panel to see which service actually caused the failure, instead of checking each log individually.

End of article