Edge Functions Down With NOT_FOUND_FUNCTION_BLOB? There's a Fix We're Validating

What's happening

Since July 31, 2026, some Lovable Cloud projects have every edge function returning:

HTTP 404 — NOT_FOUND_FUNCTION_BLOB

The pattern looks like this:

• Every edge function in the project returns the 404, not just recently changed ones
• Deployments report success, but the new bundles are immediately unavailable
• Database, Auth and Storage stay healthy
• Scheduled (cron) functions boot and shut down without producing application logs
• Everything built on edge functions goes down with them: transactional emails, SMS, AI features, scheduled jobs

The timing matches the Supabase Management API incident from July 31 to August 1. The incident was marked resolved, but some projects never recovered on their own. Five affected projects have been reported in the Lovable Discord so far.

The working theory

When a function deploys, its code is stored as a bundle (a "blob") that the edge runtime loads. In the affected projects, the function records seem to point at blobs that are gone or unreachable, and redeploying the same function keeps pointing at the broken reference. That's why redeploys "succeed" and change nothing.

Renaming the function creates a brand new function record with a fresh blob upload, so it stops pointing at the broken file entirely. One builder confirmed this fixed their project on August 1, after regular redeploys did nothing.

The workaround (test it safely first)

Don't rename everything at once. Validate with one function that doesn't matter much:

On Lovable Cloud, paste this to Lovable:

Take one non-critical edge function, rename it (for example add -v2 to
the name), update whatever calls it, and deploy only that function.
Then call it once and tell me if it returns 200 or still
NOT_FOUND_FUNCTION_BLOB. Don't touch anything else.

On your own Supabase, the same idea via CLI: copy the function folder to a new name, supabase functions deploy new-name, and call it once.

Then:

1. If the renamed function comes back alive: repeat the rename for your critical functions first (transactional email, payments, auth-related), then the rest. Your app comes back while the root cause gets fixed.
2. If it still returns the 404: stop there, don't keep redeploying. Your case has a different shape and support needs to see it as-is.

Either way, tell support

Reply to your support ticket (or open one at support@lovable.dev) and include:

• That your project never recovered after the July 31 Supabase incident
• Whether the rename test worked or not
• Your project reference and a couple of failing function names

There are multiple open reports of this being investigated internally. That context helps the team connect your case to the others instead of starting from zero.

Validation status

Hit this error? Found something different? Tell me in the Lovable Discord (community-support channel) or wherever you find me. This page gets updated as we learn.