38 QA prompts, a 15-point review checklist and a browser app that checks AI-written test work before your name goes on it.
30-day refund. Runs in your browser. Works with the AI you already use.
AI drafts. You verify.
This is a real-looking AI test summary. It reads fine. Would you sign it?
The data scan flags confidential data and blocks copying until it is removed.
Copy the prompt into your own AI and paste its answer back.
15 checks, marked by you. The automatic checks point at problems, they don't decide.
3+ fails? Regenerate rather than patch.
An evidence log for anything going to stakeholders.
Open the browser app and watch the checks run on a sample AI test summary. See it before you buy.
Open the live demoAdd the connector and the copy and paste disappears. Claude drafts, the app checks, you sign off.
Claude fetches the kit prompt you need
Claude writes the draft and sends it into a review in the app
The app runs the automatic checks
You mark the 15-point checklist. Claude can suggest, never tick
You sign off and the evidence is logged
In Claude, open Settings, then Connectors, then Add custom connector, and paste the connector address:
https://trustworthy-ai-testing-app.lovable.app/mcpNeeds an unlocked app licence (PDF kit + App, Team or Department).
Emails outside test domains, UK mobiles, Luhn-valid card numbers, NI numbers, API keys, tokens, private keys, connection strings, internal hostnames.
Figures not in your source, and recomputed pass rates.
“Low risk”, “all functionality verified”, and recommendations that differ from the one you asked for.
“Works as expected”, “behaves correctly” and other lines nobody can test.
TODO selectors, fixed waits, focused tests.
WCAG 2.2 criterion numbers to confirm, and numbers that are not valid 2.2 criteria, such as 4.1.1.
Pattern checks miss things. Check 15 is always yours.
[09:14:02] INFO login ok user=jane.doe@acme-corp.com
[09:14:03] DEBUG db connect postgres://admin:S3cr3t@db.internal
[09:14:05] WARN payment retry card=4111 1111 1111 1111
[09:14:06] DEBUG s3 upload api_key=AKIA************EXMP
[09:14:09] ERROR see ticket https://jira.internal.acmeExample evidence log
| Date | Artefact | Checks | Reviewer |
|---|---|---|---|
| 04 Nov | Release 4.2 summary | 15/15 | C. Brown |
| 03 Nov | Checkout test cases | 15/15 | C. Brown |
| 03 Nov | Bug report BR-311 | 12/15 → fixed | C. Brown |
£14
£39
Licence: Personal and commercial use (1 person; Team 5, Department 20)
30-day refund if it isn't useful. Prices shown before any VAT Gumroad adds at checkout.
Buy the PDF kit + App on Gumroad.
Your licence key is on your Gumroad receipt.
Open the app at trustworthy-ai-testing-app.lovable.app and create an account.
Paste your licence key to unlock it. Then use it with the AI you already use.
Bought the PDF kit only? Upgrade to the kit + app on Gumroad to get a licence key.
16 years in QA, most of it in regulated industries: aerospace, health and safety compliance, and SaaS.
“The drafts are fast. The risk is that nobody checks them properly.”
AI drafts. You verify.