Blog
Deep dives on the failure modes of AI-generated code — what to look for, how to fix, and how to keep vibe coding without shipping loud bugs.
The Top 5 Bugs I See in AI-Generated Code (2026 Edition)
After shipping dozens of vibe-coded apps and debugging other people's, I kept seeing the same five failure modes. Here they are, ranked by how much pain they caused.
Read more→How to Actually Review a Claude Code Diff (Without Trusting It)
You hit Accept. You scan the green. You move on. That's how the silent bugs get in. Here's a faster way to review that catches the real problems.
Read more→The Security Checks AI Coding Tools Skip by Default
Your AI assistant is optimized for 'it runs'. Security is a concern you have to bring to the conversation yourself. Here's what to explicitly ask for.
Read more→When Claude Code Gaslights You (And How to Break the Loop)
'Actually, the bug is in your environment.' 'The previous code was already correct.' If you've heard these, you're in a confident-wrong loop. Here's how to break it.
Read more→Shipping an AI Prototype to Production Without Rewriting It
The vibe-coded prototype works. Now someone wants to pay for it. Here's the minimum you need to harden before it becomes a real product.
Read more→What Vibe Coding Cannot Do (As of 2026)
After a year of trying, here are the tasks where 'just ask the AI' doesn't work. Knowing the edges saves you from prompting in circles.
Read more→Claude Code Best Practices I Actually Follow (2026)
Twelve habits I picked up from shipping code with Claude Code over the past year. None of them are 'use version control' — you already know that.
Read more→