ERR-742/Replit · Preview works, deployment fails

Appears when: Preview works, deployment fails

Replit app works in preview but fails after deployment

Preview and the published app are separate environments. If preview works but production fails, compare the deployment type, run command, production secrets, database, and live request logs before changing application code.

Reviewed October 2026 · 7 min read · By Zenveus Engineering

Follow the fix Ask an engineer

How do I fix replit app works in preview but fails after deployment?

production-fix-plan.txtrun in order

01# Preserve the working revision

Tag or duplicate the current project and record the preview flow that works.

02# Correct one production difference

Start with the first missing secret, wrong command, database mismatch, or blocked origin proved by the logs.

03# Republish and test cleanly

Use the public URL in a private window and repeat the exact acceptance flow.

04# Add a release check

Automate or document the request that distinguishes a working homepage from a working product.

Preserve the current working revision, change one proven boundary, and verify the published path in a clean session.

What if the quick fix does not work?

01Reproduce on the published URL

Use a clean browser session and record the first failed network request. A preview tab is not production evidence.

What the result tells you: A live-only failure points to deployment configuration, production data, secrets, or runtime behavior.

02Compare build and run commands

Confirm the published deployment starts the same application entry point you tested and that the selected deployment type supports the app.

What the result tells you: A successful build followed by a dead app usually moves attention to the run command, port, or server startup.

03Compare production secrets

Verify every required key exists in the published environment. Do not print secret values; compare names and availability.

What the result tells you: Undefined credentials or a production-only 401/403 means the runtime is missing or using the wrong configuration.

04Check the production data path

Confirm the deployment is connected to the intended production database and has the expected schema.

What the result tells you: An empty or incompatible production database can make a healthy UI look broken.

Likely cause	What proves it	First safe action
Missing production secrets	A live-only failure points to deployment configuration, production data, secrets, or runtime behavior.	Preserve the working revision
Wrong deployment type or command	A successful build followed by a dead app usually moves attention to the run command, port, or server startup.	Correct one production difference
Preview and production data differ	Undefined credentials or a production-only 401/403 means the runtime is missing or using the wrong configuration.	Republish and test cleanly
Production-only origin rules	An empty or incompatible production database can make a healthy UI look broken.	Add a release check

Why do AI-built apps hit this problem?

Missing production secrets

Workspace values were available in preview but were not configured for the published app.

Wrong deployment type or command

A server-backed app was published as static output, or the production start command points at the wrong entry point.

Preview and production data differ

The live app is connected to a separate, empty, or schema-incompatible database.

When the built-in AI fix makes it worse

Freeze generated changes, restore the last known working version, reproduce one request, collect the browser and platform logs, and change one layer at a time.

How do I stop this problem recurring?

→Maintain separate development and production environment manifests.
→Validate authentication, writes, and external integrations after every publish.
→Keep production configuration in an explicit release checklist, not in chat history.
→Test the published URL in a clean browser session before calling the release complete.
→Keep a last-known-good deployment and a documented rollback step.
→Log the critical request path with enough context to identify the failing layer.

Still stuck with replit app works in preview but fails after deployment?

AI prototype hardening

We restore the critical path and document the root cause.

Production failure reproduced with evidence
The smallest responsible layer repaired
Rollback and prevention guidance included

Discuss the production issue →

Replit app works in preview but fails after deployment questions

Why does Replit preview work when deployment does not?

Preview runs in the development workspace. The published app has its own runtime, URL, secrets, data, build, and start behavior.

Should I rebuild the app?

Not until the first failing production boundary is known. Most preview-versus-live failures are narrower than a rebuild.

What evidence should I send an engineer?

Send the live URL, deployment type, build and runtime logs, first failed network request, and secret names with values removed.

About the guide

Zenveus Engineering works on production systems across product, data, authentication, infrastructure, and AI integrations. This guide is based on current official platform documentation and evidence-led repair practice.