Join the Version 1 team at this Claude Code hackathon event run in partnership with Anthropic.
You’ll develop your understanding and expertise in Claude Code by building a setup around Claude Code that lets agents do real delivery work, with you stepping in only where a person adds value.
And to add a bit of fun, we’ll make it a competition. The best example will be chosen based on scoring which rewards how far your factory gets, whether it can be trusted and the evidence you show for both. Lines of code earn nothing.
Register today to book your place
How the day runs
- Build sprints with short mentor check-ins. Mentors help you unblock decisions; they don’t write code.
- Possible curveballs: facilitators announce policy changes, vulnerabilities or urgent requests. Good responses earn points.
- Side quests: optional extras for enthusiastic participants.
No dress code, just bring your laptop. We’ll provide the food, and hopefully you can join us for a couple of drinks afterwards.
Who should attend
Senior engineers and engineering leads who are familiar with AI code assistants like Claude Code and are comfortable with CI/CD, version control and automated testing.
How the build will work
Your starting point: a legacy service
Everyone inherits the same repository: the Hardship Support Payment service, an ageing Python Flask app with outdated dependencies, few tests and documentation that doesn’t match the code. Finding its hidden problems earns extra points.
Six levels
Each level builds on the last. By the end, your factory can:
- Agent-ready repo: understand the codebase well enough to get a task right first time
- Guardrails: avoid leaking secrets, logging personal data or running destructive commands
- Specialist agents: plan, build, test and review a ticket, with you approving the merge
- Orchestrated pipeline: work through the backlog unattended
- Operate: respond to a live incident and ship a fix
- Prove and package: show its reliability and install for another participant