
Can Your Agent-Written Code Survive Production? Coding agents can write impressive happy-path software. But production does not run on the happy path. Join Reboot and HackerSquad for an evening build night for engineers using Claude Code, Codex, Cursor, and other coding agents to ship real software. We will put agent-written code under production pressure and test whether it survives the kinds of bugs that ordinary demos and test suites often miss. The night is built around five failure modes: The retry that charges twice The timeout that lied The 200 response that hid a failure Two buyers, one remaining item The crash between the charge and the email You will learn how these bugs show up in agent-written applications, how to spot them in generated code, and how better system design can prevent whole classes of production failures. After the technical framing, you will build with your preferred coding agent and submit a short proof package: what you built, which failure modes you tested, what the agent initially missed, and how your final app handles production pressure. This is not a passive talk or a product pitch. It is a hands-on reliability challenge for builders who want agent-written software to be more than impressive in a demo. Who Should Apply Senior engineers and engineering leads AI engineers and product engineers using coding agents Technical founders building agent-assisted products Backend, infrastructure, fintech, marketplace, and reliability-minded engineers…
625 2nd St, San Francisco
Bright Data, 625 2nd St, San Francisco, CA 94107, USA