Monitor your Replit app in production

How a Replit app runs in production

Publishing a Replit app — what Replit used to call deploying it — runs it as an Autoscale, Reserved VM, Static or Scheduled deployment, on a replit.app address or your own domain.

Replit gives an app its own PostgreSQL database, with separate development and production copies, Replit Auth for sign-in, and a Stripe integration. The Git pane pushes to and pulls from a GitHub repository.

What breaks after launch

It works in preview and breaks in productionReplit’s own troubleshooting guide says secrets set in the workspace do not automatically carry over to the published app — the usual reason sign-in or payments work in the editor and fail live.How to tell, and the fix
Sign-in breaks on your new domainA callback or redirect URL that still names the development address fails once the app answers on its published domain.How to tell, and the fix
Payments stop after a deployA Stripe webhook pointing at the development address, or a signing secret missing in production, means payments succeed at Stripe and never reach the app.How to tell, and the fix
The site is downA published app that does not answer its health check in time never comes up; Replit’s guide gives the home page five seconds.How to tell, and the fix

What Keelnest watches

The site, every five minutes. The production address is requested the way a browser would ask for it, then the certificate and the security headers. An incident opens after two failed checks in a row, never on one blip, and a certificate is flagged from fourteen days before it expires.
Every publish. Replit does not announce a publish to anyone else, so each five-minute check fingerprints the live page. A new fingerprint is a publish: the new version is checked and your journeys are walked against it.
Sign-in and checkout, in a real browser. A journey is the steps a customer takes, written in plain words and walked in a real browser after every publish and on your plan’s schedule. It can type a one-off inbox address and wait for the email, so a sign-up confirmation or a magic link is walked to the end. A payment journey stops before the pay step; nothing is charged.
Payments. With a read-only Stripe key, payments are compared with what that hour of the week usually brings, and webhook endpoints and failed deliveries are read, so money that stops arriving is noticed in hours rather than at the month’s close. Stripe.
AI spend. With an admin key for OpenAI or Anthropic, usage is read hourly and a spike is caught within two hours, so a runaway loop shows up today rather than on next month’s invoice. Paid plans add budgets that alert at 80% and 100%. OpenAI, Anthropic.
Errors. One script tag reports uncaught errors and failed requests from real visitors — never form values or request bodies — so an incident says what broke, not only that something did. Sentry works too, if the app already has it. Error capture, Sentry.

When you publish

Within five minutes of a publish, Keelnest sees the new version on the live site and verifies it: the address, the certificate, the headers, then your journeys — which is where a secret that did not carry over shows itself, as a sign-in or a checkout that fails.

Fixes: a prompt, or a pull request

Every incident carries a prompt you can give Replit Agent, never asking for a secret. With the project connected to GitHub and a paid plan, Keelnest can prepare the fix as a pull request instead; once a named person approves it and it merges, pull it into the project from the Git pane.

Setting it up

Add the published address — replit.app or your own domain. The first check runs straight away.
Connect Stripe with a restricted key if the app takes payments, and OpenAI or Anthropic with an admin key if it calls a model.
Write one journey — sign in, or reach the checkout — in plain words. It is walked after every publish.
For fixes as pull requests, connect the project to GitHub from the Git pane and install the Keelnest GitHub App on that repository.

Or start without an account: the free health check reads any live address in a few seconds.

Questions

Does Keelnest see Replit’s deployment logs?

No. It watches what your customers see — the live site, the journeys, payments and spend — and the error-capture script reports what fails in their browsers.

How quickly does Keelnest notice a publish?

Within five minutes: the live site is fingerprinted on every check, and a change is treated as a publish and verified.

Can Keelnest fix the problem for me?

It writes the fix as a prompt on every plan. On paid plans with the project on GitHub, it can prepare a pull request instead, and nothing merges until a named person approves it.

Is it free?

Your first app is: the five-minute checks, the check after every publish, and one journey, walked weekly and after every publish. Paid plans add journeys, run them hourly or faster, and prepare fixes as pull requests.

Sources

What Replit does, as its own documentation described it on 23 September 2026. Builders change quickly; if something here is out of date, tell us.

Monitor your Replit app.