today's signal · no scroll
the brief
Agents kept inching toward autonomy and everyday usefulness: ChatGPT previewed personal finance, Vercel’s v0 learned to drive the apps it builds, and a new agent‑first language, Zero, debuted. Dev tools shipped steady polish—Claude Code added plugin safety and token‑cost visibility, Next.js refreshed its dev overlay, and Zed embraced ChatGPT subscriptions—while OpenAI probed GPT‑5.5 regressions and Erlang/OTP 29 landed.
Users report degraded GPT‑5.5 behavior; the team says systems look healthy but is investigating and will share updates—watchful users should verify critical workflows.
Codex team is aware of reports of GPT-5.5 performing worse for some users and investigating. We don't have anything conclusive yet and systems are healthy but we will share updates as we go.
Pro users in the U.S. can securely connect bank and card accounts, query spending, and get insights—another step toward always-on personal agents.
A preview for Pro users: a new personal finance experience in ChatGPT. Pro users in the U.S. can securely connect financial accounts, see where their money is going, and ask questions based on the information they choose to connect. Your full financial picture, now in ChatGPT. pic.x.com/NjbJqOqFRi
Vercel's v0 can open and operate the apps it generates, critique and debug flows, and send live screenshots, moving from codegen to full-stack UI interaction.
Introducing v0 Browser Use. v0 can now open the apps it builds, use them, critique designs, debug complex flows, and fix things proactively. While it's working, v0 will send you screenshots of what it sees. pic.x.com/lPdFSbZSDq
A new systems language with explicit capability model, JSON diagnostics, and type-safe auto-fixes aims to be easily used and repaired by software agents from day one.
Introducing Zero The programming language for agents. I wanted a systems language that was faster, smaller, and easier for agents to use and repair. Explicit capabilities. JSON diagnostics. Typed safe fixes. Made for agents on day zero. pic.x.com/uTrDOmyBR1
Release adds plugin dependency enforcement, transitive enablement, projected per-turn context costs in marketplace, and an opt-out isolation setting for background worktrees—sharpening multi-plugin workflows.
v2.1.143 — What's changed Added plugin dependency enforcement: claude plugin disable now refuses when another enabled plugin depends on the target (with a copy-pasteable disable-chain hint), and claude plugin enable force-enables transitive dependencies Added projected context cost (per-turn and per-invocation token estimates) to the /plugin marketplace browse pane Added worktree.bgIsolation: "none" setting to let background sessions edit the working copy directly without EnterWorktree, for r...
v16.3.0-canary.21 introduces a cleaner error overlay shell with instant fix-card guidance and tracing improvements, tightening the feedback loop for debugging in development.
v16.3.0-canary.21 — Core Changes Redesign dev overlay: cleaner shell + instant fix-card guidance: #93755 Misc Changes Trace middleware/proxy source files in webpack NFT pipeline: #93871 Credits Huge thanks to @unstubbable and @yavorpunchev for helping!
You can bring your ChatGPT subscription and Codex-style rate limits into Zed’s agent, signaling continued third‑party support for subscription-based access over usage‑metered APIs.
You can now use your ChatGPT subscription in the Zed agent, with the same usage and rate limits you benefit from in Codex directly. We're grateful that @openaidevs continues to support subscription-based access for third-party tools, even as others move toward usage-based pic.x.com/d11qIGsw6c
The team promises push notifications, /fork, restore after revoke, better reconnects, and improved device control as the ChatGPT mobile Codex preview iterates quickly.
Thanks for the feedback on Codex in the ChatGPT mobile app. While it’s in preview, we’re working to improve it fast. What you can expect next: push notifications, /fork, ability to restore after revoking, better reconnects, fixing the ability to control other devices, fewer
Major release of the Erlang runtime and libraries lands, keeping a cornerstone of distributed, fault-tolerant systems modernized for today’s concurrent workloads.
Erlang/OTP 29.0 — Article URL: https://www.erlang.org/news/188 Comments URL: https://news.ycombinator.com/item?id=48155297 Points: 125 # Comments: 8
A candid look at running ~100 Codex agents continuously to review and build software, reframing product design when token costs are treated as effectively free.
People freaking out over my AI spend. What nobody sees: Part of what excites me so much about working on OpenClaw is that I'm trying to answer the question: How would we build software in the future if tokens don't matter? We constant run ~100 codex in the cloud, reviewing
Matt Pocock argues long skills are hard to audit, maintain, and afford—shorter, focused skills build trust and keep token costs and complexity in check.
Long skills are such a red flag to me - Hard to audit (and therefore, trust) - Hard to edit (more text, harder to maintain) - Expensive to run (more text, more tokens) The shorter the skill, the better IMO
A glimpse at a 4,032‑GPU, liquid‑cooled Texas cluster and its scrappy origin story underlines the hardware arms race shaping modern ML training and inference.
Jane Street just showed the inside of their AI training data center in Texas. 4,032 GPUs. 56 racks. 8,000 km of fiber. liquid cooling running through every server because air cooling can't handle the heat anymore. but the part that got me was the origin story. Ron Minsky, who pic.x.com/mRchMaGskJ