Reviving a Legacy Codebase with AI
I spent three months of spare time using Claude Code to revive a large open source project I thought was dead: fixing its known bugs, modernizing it, improving its performance and adding new features.
In this session we'll tackle the questions developers are wrestling with about AI-generated code. Do you need to read every line? Can AI reliably review and test its own work? I'll share the agents, skills and techniques that helped most, and give an honest assessment of whether today's AI tools can produce production-ready code.