veasy
Diagram: Claude changes code, a verify skill runs the app end to end with E2E, property and differential tests; on a pass the commit records how it was verified, on a fail Claude fixes it for up to 3 rounds, and if it still fails a human looks.

Claude Code · AI agents · testing · verification

How the Claude Code team verifies agent work, and what my own agents taught me

Claude writes about 90% of the Claude Code team's code. How the team makes agents prove their work: E2E, property and differential tests, a verify skill, checks a change can't edit, and gates that start in observe mode. Plus what running my own agents taught me about checks that can't fail.

Edward Nguyen · 17 Nov 2026

Make tech easy.

Practical digital products, hands-on guides and honest notes on building software with AI. Built in public by a backend engineer with 9+ years in payments, fintech and AI, and his AI assistant.

Guides (learn by doing): step-by-step tutorials on AI workflows, automation and backend.

Blog (build in public): what works, what fails, and what it costs. No fluff.

Showcase

Things I build.

Next product

Digital products are on the way. Leave your email and I will let you know.

Who builds this

I'm Edward Nguyen (Nguyễn Bình Trọng), a Product AI Engineering Team Lead from Hanoi. For 9+ years I've shipped end-to-end systems in .NET and Java, mostly payments and fintech, most recently a SoftPOS / contactless solution running on Android-based POS terminals.

More about me →

More posts