GPT-5.4 vs Opus, Claude Code Loops, and 10x Engineers in 2026
And more AI Coding news: March 11, 2026.
Hey hey,
A usual weekly newsletter about AI coding with the videos and links from the community. It seems like the ecosystem slowed down a bit with new “revolutionary” releases (not for long?), and the world tries to catch up and adapt LLMs to their workflows.
For Paid Substack Subscribers
Code Quality Test: Opus 4.6 and GPT-5.4 Built a Telegram Bot with Laravel
A 13-minute video. I’ve built a Telegram bot with Laravel and Opus/Codex. Which produced a better quality code?
Will Opus/Codex Handle LEGACY Code? (No CLAUDE.md)
A 9-minute video. I tested Claude Opus and GPT-5.4 Codex on an older Laravel project called Bookstack.
My YouTube videos
I Tried New GPT-5.4 vs GPT-5.3-Codex: Is It Better?
I’ve compared two models on roughly the same prompts in Codex CLI/App and in regular/fast modes.
AI Made 10x Engineers Obsolete? My Reaction (Here’s How You Should Think)
Reacting to a video going viral on Twitter.
I Tried Codex App (GPT-5.4) to Build a Telegram Bot: Is it “Usable”?
I decided to test the Codex App (not CLI) for something more serious than just a few prompts.
NEW Instruckt: Visual Annotations for Claude Code and Other AI Agents
A new tool by Josh Cirre to avoid making screenshots for visual changes on the websites.
From AI Coding Community
Introducing Code Review, a new feature for Claude Code.
x.com
When a PR opens, Claude dispatches a team of agents to hunt for bugs. The community has a lot of tweets about it being expensive ($25 for a review) and also why Claude Code doesn’t produce the “right” code from the first attempt.
Today we’re launching local scheduled tasks in Claude Code desktop.
x.com
Create a schedule for tasks that you want to run regularly. They’ll run as long as your computer is awake.
I built a virtual design team plugin for Claude Code
reddit.com
9 roles, 16 commands, 5 agents. Hey everyone, I’ve been building Claude Code plugins and wanted to share one that’s been genuinely useful for my own workflow.
I’ve been going deep on Claude Code lately and honestly it’s been a weird experience.
Codex Update 0.110.0 is here
x.com
This release quietly introduced some big upgrades to Codex: /fast mode faster shell execution multi-agent workflow upgrades plugin + marketplace system better memory
We’re launching Codex for Open Source
x.com
Maintainers can use Codex to review code, understand large codebases, and strengthen security coverage without taking on even more invisible work.
How to enable 1m context window in Codex
x.com
Quick video tip from X.
Claude Code wiped our production database with a Terraform command.
x.com
It took down the DataTalksClub course platform and 2.5 years of submissions: homework, projects, and leaderboards. Automated snapshots were gone too.
A GitHub Issue Title Compromised 4,000 Developer Machines
grith.ai
A prompt injection in a GitHub issue triggered a chain reaction that ended with 4,000 developers getting OpenClaw installed without consent. The attack composes well-understood vulnerabilities into something new: one AI tool bootstrapping another.
That’s a wrap for this week. Keep building with AI!
Povilas Korop
AICodingDaily.com

