Science-Based Software App Audit

Your app, graded like we grade everything.

Paste one prompt into Claude Code, Codex, or Cursor. Your agent runs 38 evidence-graded checks on your actual codebase and writes a report styled like our playbooks — every finding with its file location, every recommendation with its research, and a fix plan it can implement diff by diff.

Runs in your own coding agent — your code never leaves your machine. Re-run it as often as you like.

How it works

One prompt. One session. One report.

1
Buy once, sign in

Secure Stripe checkout, $79. Your receipt email unlocks the prompt — a six-digit code, no password.

2
Paste the prompt

Open your coding agent at your project root and paste. It maps your app, asks four scoping questions, and runs the registry.

3
Read, then fix

You get sbs-audit-report.html — scoreboard, findings with file locations, and a fix plan your agent implements one diff at a time.

No invented numbers, by designThe prompt forbids guessing. Anything the agent can't verify from your code is marked UNVERIFIED in the report — stated, not styled away.
Every rule carries its researchFailing checks cite the study behind them — the 104-person eye-tracking result, the 9.2mm thumb-error floor, the randomized 337k-user trial experiment.
Conventions labeled as conventionsWhere a rule is our judgment rather than an experiment — one dominant CTA, three tiers — the report says so. That's the whole brand.
What gets checked

38 checks across nine areas.

Distilled from the Science-Based Software playbooks and our graded research library — 690 claims, 559 traced sources. Sections that don't apply to your app are marked N/A, never guessed.

4TYPTypography & contrast

16px floors, real WCAG math on your actual hex values, dark-mode pairs flagged for human eyes.

4TAPTouch targets & labels

48dp hit areas, 8dp spacing, and no icon-only controls on the paths that matter.

4CONConsistency

One term, one button style, one interaction per concept — plus a generated pattern sheet.

4HIERHierarchy & layout

One dominant action per screen, shared alignment edges, intentional grouping.

3DISCProgressive disclosure

A reduced first-run, labeled advanced levels, and the reveal path that repays what you hid.

5FRUNFirst run

Value before the signup wall, one question per screen, primed permissions, honest progress.

5TRIALFree trial

Length, charge timeline, defaults, cancel-anytime — and no countdown sharing the screen.

5PRICEPricing page

Tier count, interior highlight, per-month framing, zero urgency theater, a real 9-ending policy.

4TRUSTTrust surfaces

No unsourceable stats in your copy, renewal terms up front, a two-tap cancel, no dead-end errors.

What this is not

It is not a human consultation, and it will not promise a conversion lift. It is the Science-Based Software evidence registry, encoded so your own agent can hold your app against it — including where the evidence is telemetry with limits, and where a rule is our labeled judgment. Taste (grids, typeface pairings, palettes) is graded as taste in our library, so the audit leaves it alone.

Which tools does it work with?
Any agent that can read your codebase and write files: Claude Code, Codex, Cursor, Windsurf, and similar. The prompt is plain text — no integration, no API keys, nothing to install.
Do you see my code?
No. The audit runs entirely inside your own coding agent on your machine. We never receive your code, your report, or anything else.
How long does it take?
One session. You spend about five minutes answering scoping questions; the agent does the rest. Implementing fixes is up to you — each ships as a reviewed diff.
Is this a subscription?
No. $79 once. The prompt is yours to re-run before every launch, after every redesign, forever — including registry updates we publish.
Refunds?
One email within 14 days, full refund, no questions. Removing purchase fear in writing is our own research talking.

Your App Audit

·
1. Open Claude Code, Codex, or Cursor at your project root. 2. Copy the prompt below and paste the whole thing — registry, template, and all. 3. Answer the four scoping questions, then read sbs-audit-report.html when it lands.
THE Science-Based Software APP AUDIT · PASTE INTO YOUR CODING AGENT
Loading your prompt…