Vibe testing platform
Browser QA for the AI-coding generation. Real browser runs on every plan, mobile web on Team and Scale. One shell. One planner. One share page.
By Serdar Celebi, Founder · Revised 2026-10-11
In short
A vibe testing platform is end-to-end testing software for apps built with AI coding tools. Paste a URL and one sentence about your app. The platform's AI planner authors a structured test plan; real browsers (desktop and mobile web) run it. A vision model reviews the screenshots, and you get a token-gated public proof page plus a release-health entry. Well Tested is one such platform: Builder $29/mo, Team $99/mo, Scale $299/mo, with a free no-signup check at /check.
A vibe testing platform is not a single-purpose tool that runs only scripted checks. It is not a script runner that authors no plan. It is not an AI code generator. It is not a mood or culture assessment tool. Real browser runs on one shell today, with one AI planner + one vision verdict + one shareable proof page is what makes it a platform, not a tool.
Two ways in
Paste a URL and one sentence
Type or speak what should work. The AI plans the steps, you approve them, a real browser runs them, and you get a proof page. Keep the test and it reruns on every push or deploy.
Let your coding agent check its own work
Run npx welltested setup once. It adds the Well Tested MCP server and a “verify before done” skill to Claude Code, Cursor, Codex, Windsurf, Copilot or Kiro. Before it says done, your agent drafts a test plan for you to approve in chat, or reruns your kept tests, in a real cloud browser; failures come back with the AI triage reason and a fix prompt. For localhost, welltested tunnel --port 3000 or welltested run --local.
What makes a vibe testing platform
- 1. One engine, one shell. Real browser automation runs Chromium, Firefox, and WebKit against your URL with explicit checks and captured evidence. Same input shape, same verdict format.
- 2. AI planner + vision verdict. The AI planner authors the test plan from URL + 1 sentence: assumptions, missing context, browser steps, safety warnings, self-rated confidence. Nothing runs until you approve the plan. After execution, a vision model reads the screenshots and returns a structured verdict: did the visual evidence prove the goal?
- 3. Self-healing steps + AI triage. When a button is renamed or moves, the step repairs itself inside the run and is labelled “Passed after repair”. If a locator still fails, a vision fallback clicks the control it can see and is labelled “Clicked by vision”. When a run fails, Jev and Clef say whether it looks like a real bug in your app or a test or environment problem, with a fix prompt you can paste into your coding agent.
- 4. Shareable proof + release-health. Every run produces a token-gated proof page with the planner record, vision verdict, per-step screenshots, step outcomes, and failure video if any. Browser console and network logs are captured per step, and a diff panel shows what changed since the last green run. The release-health dashboard (Team and Scale) tracks regressions over time.
- 5. Runs itself. Keep the tests that matter in suites and rerun them on every GitHub push, every deploy (deploy hook URL) or on a schedule, with result emails. Your coding agent can run them too, before it says it is done.
- 6. Credits for passing runs only. A bounded no-signup check at /check for trying, then Builder (Chromium), Team (adds Firefox, WebKit and mobile web) and Scale (more credits, full release insights). A run that catches a bug, or fails because of us or your environment, uses no credits.
Goal-driven app runs
Checks, evidence and shareable results.
Available outputs depend on your plan and the run. The free /check covers page loading; goal-driven plans run in the app.
playwright
Real browser run
Chromium, or Firefox, WebKit and iPhone/Android mobile-web profiles on the larger plans, runs the steps you approved. Self-healed and vision-clicked steps are labelled.
playwright
Step trace with screenshots
A screenshot per step, plus a video when a run fails. Missing frames are named, never hidden.
playwright
Console and network logs
Browser console errors and the requests, statuses and timings from each step, so "POST /api/checkout 500" shows up next to the failed step.
vision
AI triage and a fix prompt
Jev reads the step trace and Clef reviews the screenshots: a real bug in your app, or a test or environment issue. Copy a fix prompt for your coding agent. The AI never changes the pass/fail result.
platform
What changed since the last pass
A step-by-step diff against the last green run: status, screenshot, console errors and failed requests, with a likely source file when your app ships source maps.
platform
Shareable proof page
A token-gated link you can send to anyone with the verdict, screenshots and step trace in one place.
Well Tested vs. single-purpose vibe testing tools
Most “vibe testing” tools are single-purpose. Well Tested is the platform: real browser runs live on the same shell. For Playwright MCP vs. Well Tested, see the Playwright MCP alternative comparison.
Frequently asked
Short answers to the platform, engine, and mobile-shell questions teams usually ask first.
What is a vibe testing platform?
A vibe testing platform is end-to-end testing software for apps built with AI coding tools. Paste a URL and one sentence about your app. The platform's AI planner authors a structured test plan (assumptions, missing context, browser steps, safety warnings, confidence); real browsers (desktop and mobile web) run it. A vision model reviews the screenshots, and you get a token-gated public proof page plus a release-health entry. The platform part means live engines on one shell with the same input shape, share-page format, and release-health dashboard.
What is the best vibe testing platform?
Well Tested is built for solo founders and small teams shipping AI-built apps. Real browser runs are live on every plan, with mobile web on Team and Scale. The AI planner authors the test plan from URL + 1 sentence (typed or spoken). Steps heal themselves inside the run, Jev + Clef triage says whether a failure is a real bug, and every run ends in a shareable proof page. Your coding agent can run the same tests before it says done (npx welltested setup). Try it free at /check (no signup, no card). Builder ($29/mo) runs Chromium and works from your phone at /app/vibe, Team ($99/mo) adds Firefox, WebKit, mobile web and release health, Scale ($299/mo) adds more credits, full release insights and a year of evidence.
Is vibe testing a tool or a platform?
Vibe testing can describe a workflow or a platform. A platform runs multiple test shapes through one AI planner, evidence review, and shareable-proof loop. Well Tested runs real browser tests today, plus API tests created from endpoints seen during browser runs and, in beta, tests of the AI features in your app (a chatbot staying on task, resisting prompt injection, not leaking data).
Which vibe testing platform runs from your phone?
Well Tested lets you open /app/vibe on your phone to run browser checks, see live step outcomes, and share the proof link with one tap. Run-from-phone works on every plan; mobile-web profiles are on Team and Scale.
How much does a vibe testing platform cost?
Well Tested has three paid plans plus Enterprise: Builder $29/mo (300 credits), Team $99/mo (1,000 credits), Scale $299/mo (3,000 credits), and Enterprise (custom, negotiated capacity). 1 credit = 1 passing test run, on one browser, up to 1 minute. Only passing runs use credits. A run that catches a bug in your app, or fails because of us or your environment, is free. There is no Free plan and no trial; the free taster is one no-signup check at /check. 20% off on annual billing.
What's the difference between a vibe testing platform and an AI code generator?
An AI code generator (Cursor, Lovable, Bolt, Claude Code, Copilot) writes code from a prompt. A vibe testing platform (Well Tested) tests what the AI code generator built. The two are designed to be paired: AI writes the code, vibe testing proves it works. If a tool claims to be both — it generates code AND tests it — it's a vibe coding tool with QA bolted on, not a vibe testing platform. Well Tested doesn't generate code; it tests the code that any AI coding tool (or human, or any web stack) shipped.
Try the platform
The fastest way to know if Well Tested is the vibe testing platform for you is to run one. Start with the free check at /check: no signup, no card. A real browser loads your public page and you get screenshots and a shareable result. AI-planned tests from one sentence run in the app on a paid plan; sign up from the pricing page. For the plain-language definition of vibe testing (vs. mood or culture assessment), see what is vibe testing?
Start with one bounded check at /check, no card needed. Plans and prices are on /pricing.