Qheck

The AI QA agent that audits your app while you build.

Qheck watches your web application like a real user, captures every interaction, and writes you a prioritized Qheck Deck — findings, fix recommendations, and ready-to-paste prompts for whatever AI coder you use, or your dev team.

Join the Waitlist See how it works →

Three modes, one workflow.

Capture what you do, let the AI do it for you, or combine both. Every session produces the same downloadable Qheck Deck.

W

Watch Mode

You drive. Click around your app and mark up issues directly on the page with the built-in pen. Every annotation becomes an evidence anchor the audit cites back to.

AI Quick Scan

The AI drives. Audits the current page plus one nearby section in 1–2 minutes — no clicking required. Best for spot-checking after an AI edit lands.

🤖

AI Deep Scan

The AI drives at depth. Walks every nav section of your site. Resumes across sessions when the site is too big for one run.

Settings — Your App yourapp.com/settings S Account settings Update your profile and notification preferences. DISPLAY NAME Scott S. NOTIFICATION FREQUENCY Weekly digest Save Save button reads as disabled — same gray as the borders

You circle. Qheck cites. Your markup — circles, arrows, sticky notes — carries straight into the Qheck Deck as anchored evidence, referenced by location and your exact words.

Five focuses. Pick what you want audited.

Choose a focus per session. Qheck targets the audit, ranks findings against your goal, and surfaces anything outside the focus separately so coverage doesn't suffer.

Visual / UI

Color, contrast, typography, layout integrity. Strict — every claim grounded in computed styles and screenshots.

Functional / Bugs

Broken interactions, validation gaps, dead links, console errors. Verified against the events log before anything ships as a finding.

Friction / Workflow

Slow flows, dead clicks, redundant steps, confusing CTAs. The journey, not just the screen.

Full Analysis

All lenses at once. Heavier audit, broader coverage, dedup across categories. Use before launches.

?

Open Feedback

Looser rails. The AI is allowed to speculate — every speculative finding tagged [inferred]. Use when you want fresh eyes, not a checklist.

Optional query declaration per session adds a soft prompt — "I'm worried about checkout" — that ranks findings against your concern without filtering out coverage.

Built for the way you actually ship.

Three high-leverage moments where Qheck saves real time.

After AI Edits

Verify before you commit.

Run a 90-second AI Quick Scan after your AI coding tool finishes. Catch the dropdown that broke, the contrast that drifted, the link that no longer works — before customers do.

Pre-Deploy

Regression sweep before release.

AI Deep Scan walks every section of your app. Multi-session resume means even large apps get full coverage. Find the things you forgot to test.

New Features

Qheck what you just built.

Pick a focus — visual, functional, friction — and Qheck targets the audit accordingly. Catch issues before the field worker, contractor, or customer opens your app.

What's in a Qheck Deck.

Every session generates a downloadable .zip with everything you need to act on findings.

Join the waitlist.

Pre-release access is rolling out to a small group of early users. Add your email below.

By joining the waitlist you agree to our Privacy Policy. We'll email you when access opens — no spam, ever.

Honest answers.

Does Qheck send my data anywhere?

Yes. Screenshots, DOM data, and console output are sent to Anthropic's API for AI analysis. Today, your Qheck Deck is stored on your device. Cloud sync, team accounts, and account features are on the roadmap — when those ship, you'll opt in per session and we'll update this page. Don't audit pages that contain sensitive personal information. Read the full privacy policy.

Can the AI hallucinate findings?

It can. We've built 25+ anti-hallucination rules into the synthesis layer — numeric grounding, computed-style verification for color claims, evidence-anchored absence claims, click-count cross-checking against the events log, standards-citation discipline, and a confidence/severity coupling so low-confidence findings can't escalate to high-severity prompts. Anything we can't fully ground gets tagged [inferred] or [cannot fully verify] and flagged for your judgment. Verify before acting on those.

What does Qheck NOT find?

Backend bugs, database integrity, and security vulnerabilities outside the rendered DOM — Qheck audits what the browser shows, not what your server does. Network requests, performance metrics, console output, and runtime exceptions with stack traces are shipping today in Dev Mode (Pro tier, via Chrome's debugger). Accessibility tree analysis, mobile viewport emulation, and code coverage are coming next in v0.4 Phase 4.4.

Will it slow down my app?

No. Qheck captures screenshots and DOM data via standard browser APIs. There's no proxy, no instrumentation, no server-side hook. The app you're auditing runs at full speed.

Why is it called Qheck?

Q for the framework: Query (what to look for), Quantify (measure what's there), Qualify (judge what matters). Pronounced "check."