chekd

local-first · learns your taste

Walk your own app.
Catch what your AI missed.

The verification cockpit for apps your AI wrote.

$ npx chekd

See how it works How it talks →

Paste one line into your AI setup. Minutes later you're walking your app.

localhost:5173 / checkout

walking

brokenbtn.checkout

“Pay now” does nothing on click — handler never bound.

opencontrast 2.9:1

Muted label fails AA on this surface. Bump one step?

resolvedhero spacing

Tightened to your rhythm — matched last 3 accepts.

learning your taste

sharper every run

01How it works

Paste one line. Walk your app. Catch what slipped through.

chekd is an Iron-Man suit for your taste, not a robot that grades you. It walks the app your AI wrote, surfaces dead, broken, and inaccessible findings as correctable cards, and learns what you accept — so every run is sharper than the last.

Step 01

Paste one line

Drop npx chekd into your AI setup or your terminal. No account, no upload, no config to wade through.

Step 02

Walk your app

chekd opens the cockpit and walks your running app, scanning every control and surface as you go.

Step 03

Confirm & dispatch

Each finding is a card you confirm, skip, or hand off as a fix. You stay in the chair the whole time.

$ npx chekd

⟨✓⟩ chekd · the verification cockpit

walking localhost:5173 …

• broken checkout button has no handler

• open label contrast 2.9:1 (AA fail)

• resolved hero spacing — matched your taste

⟨✓⟩ 3 cards ready · chekd review

02Principles

Built to be yours — and to stay yours.

Three commitments shape every part of chekd, from the first run to the thousandth.

Accessible to all

Calm and welcoming. Built for non-technical builders too — plain language, never gatekeeping, three flavors in dark and light.

Local-first

Nothing uploads. Your app, your code, and your findings stay on your machine. Privacy mode keeps it fully local.

Compounding

It learns your taste, run over run. Accept, skip, or correct a finding and the next walk is already sharper.

03How it talks

A trusted copilot, never a critic.

Findings are suggestions to confirm, never verdicts. Plain language, encouraging, honest about limits — and it never judges the work you made.

✓ What it says

The “Pay now” button doesn’t fire — want me to wire it up?Concrete, names the spot, hands the decision back.
This label is hard to read on that surface. Bump it one step?Plain language a non-technical builder gets.
Matched your last few accepts — tightened the spacing.Shows it’s learning you, run over run.
I might be wrong here — skip if this is intentional.Honest about limits; never authoritative.

✗ What it never says

Critical accessibility violation detected. Fix immediately.Alarmist, gatekeeping, treats a finding as a verdict.
Your design is bad. Here’s what a real designer would do.Judges the creative work. Never do this.
17 issues found across 4 severity tiers (see dashboard).Noisy, jargon-heavy, dumps work on the user.
AI-powered QA automatically fixed your app.Overrides the builder; hype, not honesty.

Walk your own app.Catch what your AI missed.