As we all try to figure out what Mythos means for us down the line, the world of practical agentic coding continues, with the latest array of upgrades.
"Hey Claude, please look through my blog posts from the past couple of months for a list of all published Claude Code features and please give me an infographic using Nano-Banana image-gen listing them all and what they do, and which ones are currently Mac only or have other limitations like "enterprise only." Rank them in descending order by usefulness. I want to add the image to my recent Claude substack post. "
Cool stuff. It’s a shame that you can’t really try anything on the Pro tier now since the limits there have been just a bit more generous than on the Free tier for a while.
P.S. I wonder why this ongoing issue is constantly being ignored on the blog.
Already did. I use them in parallel: Codex for some long and heavyweight tasks and Claude for something more like of a one-shot type (with varying success still). I also use Antigravity but their interface is a bit clunky for me.
Auto mode shifts the value question from "can the agent do this" to "should I review this." That's the harder problem. Most teams aren't failing at automation — they're failing at maintaining enough context to catch when the agent went off-track three steps ago. The review discipline is the missing half of the agentic coding story.
Podcast episode for this post:
https://dwatvpodcast.substack.com/p/claude-code-codex-and-agentic-coding
Hi Zvi - Run this prompt for us please?
"Hey Claude, please look through my blog posts from the past couple of months for a list of all published Claude Code features and please give me an infographic using Nano-Banana image-gen listing them all and what they do, and which ones are currently Mac only or have other limitations like "enterprise only." Rank them in descending order by usefulness. I want to add the image to my recent Claude substack post. "
Cool stuff. It’s a shame that you can’t really try anything on the Pro tier now since the limits there have been just a bit more generous than on the Free tier for a while.
P.S. I wonder why this ongoing issue is constantly being ignored on the blog.
Or switch to Codex? Way more usage @ $20/mo
Already did. I use them in parallel: Codex for some long and heavyweight tasks and Claude for something more like of a one-shot type (with varying success still). I also use Antigravity but their interface is a bit clunky for me.
Auto mode shifts the value question from "can the agent do this" to "should I review this." That's the harder problem. Most teams aren't failing at automation — they're failing at maintaining enough context to catch when the agent went off-track three steps ago. The review discipline is the missing half of the agentic coding story.