For Replit apps & Replit Agent builds

E2E + Visual Testing for Replit Apps.
Replit shipped it fast. Lastest keeps it shipping.

Replit Agent generates a working app in minutes. Then you tweak a prompt, redeploy, and the signup flow is silently broken. Lastest gives you deterministic E2E + visual regression tests for any Replit app - point it at the deployed Repl URL, no local clone, no source access required.

Start Free →View source
Point at any Repl URLNo local clone needed$0 self-hosted
Lastest building a test suite for a deployed Replit app
No Clone. Just Test.

Why Not Just Ask Replit Agent to Test It?

You can ask Replit Agent to test what it built. Here's why that breaks.

Replit Agent is great at writing your app. It's not great at guarding it. Env vars break between runs. The agent loops on errors instead of converging. Deploy to prod is one click - and breaks in one click. Two approaches, and which one actually holds.

01

Ask Replit Agent to test its own work.

You ask the agent to verify the signup flow. It writes a script, hits an error in Replit DB, loops three times trying to “fix” it, eats your credits, and declares it works. Different selectors next run. No screenshots. No history. And the env vars it relied on already drifted.

  • Agent loops on errors - burns credits without converging
  • Env vars drift - what passed yesterday fails today
  • Non-deterministic - different path every run, no diff history
  • One-click deploy + one-click breakage = no safety net
02

Use Lastest. Point at the deployed Repl. Done.

Paste your your-app.replit.app URL into Lastest. Click through signup, the dashboard, and the Replit DB-backed flow you care about. AI generates deterministic Playwright with multi-selector fallback, captures a visual baseline, and replays it on every redeploy - zero tokens, zero clones, zero source access needed.

  • Works on any deployed Repl URL - no clone, no source access
  • Deterministic replays - same code, same screenshots, every run
  • 3 visual diff engines - pixelmatch, SSIM, Butteraugli
  • Self-healing 7-layer selectors - survive Replit Agent re-renders
  • Zero-token replays - replay thousands of times a day for $0

Built for the Replit Workflow

Prompt. Deploy. Then guard it.

Two features that matter most when Replit Agent is shipping new versions on every prompt.

01 · Point & Record

Paste your Repl URL. Click your flow. That's the test.

Open the recorder, paste https://your-app.replit.app, walk through signup → Replit DB-backed dashboard → checkout, hit stop. Lastest captures every click, type, and navigation and generates deterministic Playwright code. No local clone. No source access. No SSH into the Repl.

  • Works directly against the deployed Repl URL
  • Captures multi-step screenshots per run
  • No access to your Replit source needed - tests the running app
Recorder capturing a Replit app user flow

02 · Visual diff

Did the last Replit Agent prompt quietly break the layout?

Replit Agent regenerates whole files on every prompt. New markup, new class names, different wrappers. DOM tests pass and the dashboard is still 12px off. Lastest compares screenshots with three diff engines (pixelmatch, SSIM, Butteraugli) so visual drift shows up where you can actually see it - before it hits your one-click deploy.

  • Text-region-aware diffs ignore dynamic content from Replit DB
  • Branch-to-branch compare for pre-deploy review
  • Approve or reject with a one-click slider
Visual diff slider comparing baseline vs new Replit app screenshot

Common questions for Replit apps

Does Lastest work with Replit Agent-generated apps?
Yes. Replit Agent regenerates code on each prompt; Lastest tests the running app, not the source, so prompt-driven changes show up as visual diffs against the baseline. Re-prompt the Agent freely without re-recording.
Can I point it at a Replit deployment URL?
Yes. Any HTTP URL works: a custom domain, a replit.app URL, or a forwarded port. The recording follows the running app wherever it lives.
What about Replit Database, ObjectStorage, and Auth?
All exercised through the running UI. Lastest records the flow that writes to the DB, uploads to ObjectStorage, or signs in via Replit Auth, and asserts on the resulting screenshot. Backend regressions surface as visible diffs.
Does it survive Replit's hot reload?
Recordings happen in a stable browser tab Lastest controls, so hot reloads in your editor don't affect the test runner. CI replays run against a built deployment, not the editor.
Will it work with Replit Bounties hand-offs?
Yes. Once the Bounty hunter ships, record the new flow once and lock it down. Future bounties get tested against the locked baseline before merge.
What does it cost?
Self-hosted is free forever (open source, FSL-1.1-ALv2): your infra, unlimited screenshots and replays. Lastest Cloud is a flat $299 per month with no per-seat or per-screenshot fees, plus a free evaluation tier.

Add E2E + visual tests
to your Replit app.
In five minutes.

No cloud account. No API key. No procurement. Point it at your deployed Repl URL and go. Open source, FSL-1.1-ALv2 licensed, $0 forever.

★ Star & Install from GitHub →
install

# 1. clone

$ git clone https://github.com/las-team/lastest

 

# 2. start

$ cd lastest && docker-compose up -d

 

# 3. point at your deployed Repl

$ open http://localhost:3000

 

✓ View the source