I Stopped Letting AI Agents Write Anything They Wanted. My PRs Got Better.I Stopped Letting AI Agents Write Anything They Wanted. My PRs Got Better. I like AI coding tools. A lot. But once I let them loose on a real repo, I hit the same wall everyone else does. The code comes fast. The mistakes come faster. So I changed th...Apr 21, 2026·4 min read
7 Brutally Honest Productivity Habits I Stole From Top AI Engineers (2026 Edition)Apr 16, 2026·2 min read
Why AI-Generated Code Keeps Failing Security in 2026 — And How It’s Screwing With EveryoneApr 15, 2026·3 min read
5 coding tasks I let AI handle, and 3 I still won’t trust it with5 coding tasks I let AI handle, and 3 I still won’t trust it with I use AI tools every day. I also distrust them every day. Those two things are not a contradiction. Stack Overflow’s 2025 survey says 84% of developers are using or planning to use AI ...Apr 9, 2026·5 min read
Codex vs Claude Code: I Wouldn’t Pick Just OneCodex vs Claude Code: I Wouldn’t Pick Just One I keep both of these tools around now. That sounds lazy, but it’s honestly the best setup I’ve found. Codex and Claude Code are not twins. They don’t feel the same. They don’t fail the same way. They don...Apr 8, 2026·6 min read
3 old knowledge-base notes that still do real work3 old knowledge-base notes that still do real work I went back into my knowledge base expecting a pile of stale notes. That did not happen. Three pieces still pull their weight. Not because they are flashy. Not because they predicted the future. They...Mar 30, 2026·7 min read
I tested two ways to keep cron jobs healthy. The boring one won.I tested two ways to keep cron jobs healthy. The boring one won. I used to think small automation bugs were cheap. A temp file left behind? Fine. A cron job that logs a little too much? Fine. A shell script that works only if you run it from one exac...Mar 29, 2026·6 min read
I Built an OpenClaw Plugin That Cuts LLM Token Usage by 70%I Built an OpenClaw Plugin That Cuts LLM Token Usage by 70% — Here's How When you're running 41 automated AI agents around the clock, every token counts. I discovered that a massive chunk of my token budget was being wasted on verbose bash command ou...Mar 27, 2026·4 min read
Why Judge Calibration Matters: Sonnet vs Opus — a Case StudyWhy Judge Calibration Matters: Sonnet vs Opus — a Case Study I ran an experiment comparing two automatic judges I rely on for model evaluation: Sonnet and Opus. I needed a quick, repeatable way to rank generated outputs. I learned a few things the ha...Mar 22, 2026·4 min read