Test-Driven Agentic Behaviours

For this online event, we will have the opportunity to receive Antony Marcano as a speaker.
After a brief introduction of our community, Antony will deep dive through a live demo in Test-Driven Agentic Behaviours (TDAB), a test-first approach to improving AI coding agents by evolving their guidance one automated behaviour at a time.

Abstract:
What if you applied Test-Driven Development (TDD) not just to code, but to the behaviour of AI coding agents? In this session, Antony Marcano explores what happens when agent guidance and skills are treated as something that can be specified, tested, and evolved through automated examples.

Starting with a clear view of what Test-Driven Development (TDD) and Behaviour-Driven Development (BDD) actually are — and common misconceptions about both — he shows how these ideas extend naturally to agent files such as skills and guidance.

He introduces Test-Driven Agentic Behaviours (TDAB), a test-first approach to improving AI coding agents by evolving their guidance one automated behaviour at a time. Through a live demo, he shows how agent behaviours can be driven by failing scenarios, refined incrementally, and protected against regression.

Along the way, he shares some unexpected findings — including how “checkpoints”, intended to enforce standards through iteration, instead changed agent behaviour entirely. Rather than failing and improving, agents learned to avoid the cost of being wrong, often producing the correct result first time.

The session includes a live demonstration using Claude Code agents, showing the approach in practice.

About Antony:
Antony Marcano has over 30 years in software development, with more than 25 years in agile software development. He works as a consultant and coach, and has taken on interim Head of Engineering and fractional CTO roles.

In that time, he has helped organisations achieve the “Elite” DORA software delivery performance benchmark, both before and during the rise of AI-augmented coding.

Known for the Screenplay Pattern for automated acceptance tests, he more recently created Test-Driven Agentic Behaviours — a test-driven approach to improving AI coding agents one automated behaviour at a time.

His work is reflected in contributions to, and acknowledgements in, books including Agile Coaching, Bridging the Communication Gap, Software Craftsmanship Apprenticeship Patterns, and Agile Testing. He has also written for Better Software Magazine, where he served as Technical Editor.

He remains hands-on in the code, speaks at international conferences, and has been an invited speaker and guest lecturer at Oxford University, Cambridge University, and McGill University.