How I Trust Coding Agents With Ambitious Features?
Registration
Past Event
About Event
I tried handing an ambitious feature to a coding-agent loop. The feature worked. Every test passed. It also quietly wrecked my architecture.
So in this stream, I’ll show you the workflow I built instead—the process I now use with Claude Code and Codex to ship bigger features reliably.
We’ll answer:
Why a feature can work, pass every test but still wreck your codebase?
Why detailed plans and reviews aren't enough for ambitious work?
How I build a system to prompt my agents on my behalf?
If you’re experimenting with coding agents and trying to trust them with bigger features, I’d love to have you there.