Claude Code, Codex and Agentic Coding #8
When I started this series, everyone was going crazy for coding agents. Now a lot more people are going crazy for coding agents, as well they should given how much better coding agents keep getting, but also Everybody Knows they are good and is focusing on actually using them. With the slower pace of news here it’s no longer clear that the waits associated with doing these updates on their own are worthwhile, so I’m going to fold these updates into the weekly again for now unless there’s a new major development. Table of Contents Whoops, Sorry. Huh, Upgrades. Codex of Ultimate Computer Use. Rookie Numbers. I See What You Did There. Just a Ride. They Didn’t Want Our Jobs. Skilling Up. The Lighter Side. Whoops, Sorry Claude Code suffered in April from three distinct issues that have now been fixed. Default reasoning was changed from high to medium to deal with latency, but users disliked this and blamed it on the model. It was introduced on March 4 and reverted on April 7. A bug made it so that if a session was idle for an hour, older thinking would be stripped out after each future turn, not only the one time it was idle. This was introduced on March 26 and fixed on April 10. A system prompt instruction change, intended to reduce verbosity, hurt coding quality. This was introduced on April 16 and reverted on April 20. They promise to have a larger internal test of future changes before wide deployment, to prevent such issues in the future, and added some other controls. This is the flip side of moving so quickly. You’re going to make mistakes. It does seem like Anthropic got overly aggressive if there were three such incidents within a month. Huh, Upgrades Codex now has auto-review, their version of auto mode. Codex gets a major upgrade that substantially speeds up computer use. Codex adds support for 90+ plugins to give it access to all your existing tools. Codex can now work directly in Chrome via a plugin in the Codex app, and do your repetitive browser work while