engineering @ ShopBack; creator of https://xitter.cf/t.co/HI1XYMVpSN
Stockholm, Sweden
Joined August 2013
- Tweets2.5K
- Following493
- Followers616
- Likes1.7K
I'm running Claude Code, Codex, and Gemini CLI in parallel.
The worst part isn't the token cost. It's the tab switching.
AI DevKit's Agent Console fixes this. One view, all your agents, send messages without leaving, quick hotkey to focus when you need detail.
Works in iTerm2, Ghostty, your default macOS terminal. No new app.
npx ai-devkit@latest agent console
I get this. I spent a few months struggling with the same feeling. Seeing machines handle things I’d spent years learning and doing was honestly depressing. I kept asking myself what would make me relevant anymore. I still don’t have a settled answer.
But I’ve also started to see the other side. I can learn faster, explore several ideas at once, and build things I previously couldn’t even attempt. Things that used to take a whole team are now within reach of a few people. That has made me curious about building again.
I keep reminding myself that execution is part of our job, but it isn’t the whole job. We build things to solve problems for people. Understanding what someone is struggling with, deciding what’s worth building, and caring about how it fits into their life still matter. To me, that’s the human soul of a product: someone cared enough to make it useful.
I can’t promise AI will never do that too. If it eventually understands us and brings that same care to what it builds, then we’ll have much bigger questions to work through as humanity. That isn’t something software engineers alone should have to figure out.
None of this makes the uncertainty disappear. But being able to build things I care about, which used to feel out of reach, gives me something to look forward to.
Hoang Nguyen retweeted
During this red-teaming exercise, we also found a vulnerability that allows for unauthenticated remote code execution in Brotli, a widely-deployed compression library included in many pieces of software such as Node.js, Bun, and Envoy.
The flaw could let an attacker run their own code on affected systems by sending or replying with malicious compressed data, without any need to create a user account. That is, an unpatched server that accepts compressed input could be taken over by a malicious client, and an unpatched client that accepts compressed input (such as “curl --compressed”) could be taken over by a malicious server.
This root cause issue was actually fixed in the Brotli source code in February, but it wasn’t known at the time that it gave rise to a security vulnerability and there was no more recent release of Brotli. Since Brotli is built into other software, any software which includes this library needs to take the update too, followed by the users of that software. We disclosed the vulnerability to Brotli and other downstream projects such as Node.js, which released or are currently releasing security updates, and Brotli cut new release versions and published a CVE.
More details on the patch and mitigations below:
patchtheplanet.org/brotli
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
the benefit when you have full ownership is that you can ask your AI agents to try whatever you want 🎃 🎃 🎃
Hoang Nguyen retweeted
After 11 years, I'm leaving WordPress.
I paid a few bucks every month so I could focus on writing instead of dealing with hosting, setup, and maintenance.
Now my coding agent handles all of that for me.
The reason I chose WordPress no longer exists.
Hoang Nguyen retweeted
Introducing tsrs: a Rust port of the TypeScript compiler 🦀
We are not forking TypeScript!
◆ In close collaborating with the TypeScript team
◆ Up to 8x faster than TS 7.0 & ~70% less memory
◆ Runs in WASM, so it works in your browser & on Workers
◆ Soon part of the Oxc toolchain
npx tsrs
blog.cloudflare.com/typescri…
After 11 years, I'm leaving WordPress.
I paid a few bucks every month so I could focus on writing instead of dealing with hosting, setup, and maintenance.
Now my coding agent handles all of that for me.
The reason I chose WordPress no longer exists.
"Own the Craft" means achieving deep mastery, total confidence, and personal accountability over your specialized skill or profession. It means taking full responsibility for the quality of your work and how you execute it.
I told OpenCode to hire Pi to do the job and it did.
> It spun up two more agents on its own: a tests agent and a docs agent. It handed each one a task, waited for them to finish, and reported back.
> 6/6 tests pass, README updated. I didn't touch either worker.
> That's a quick demo of `ai-devkit agent start` + `agent send`, called by the agent itself.
Open source, runs locally 👇
