Hi, and welcome to this week's edition of the newsletter! As always, feel free to reply to this email to get in touch. Or ping me on any of my socials, which can be found on my personal website, danclarke.com.

If you get value from this newsletter, please help me out by sharing it on your socials and with your tech friends and colleagues - it really helps me spread the word and keep the newsletter going 🙏 (at-mention me, and I'll repost).

📌 News item of the week

Cursor and Anthropic have both shipped a feature called Projects, a week apart, and underneath the shared name they are close to the same idea.

Cursor launched Projects on September 10. Anthropic rebuilt Projects on September 17, with the new version landing first in Claude Code. In both, you stop talking to one agent that writes code and start talking to a coordinator that scopes the request, splits it into pieces, and hands those out to other agents working in parallel. Cursor's version delegates to subagents rather than writing any code itself. Anthropic's runs each piece as a separate Claude Code cloud session on its own branch and copy of the repo, and if two threads touch the same code the overlap turns up as a merge conflict, just like any other PR.

In both, the work carries on after you close the laptop, because it is running in the cloud rather than on your machine. Cursor also spins up a local agent when something needs testing on your own machine.

Both are in beta, though. Cursor's is rolling out to all users. Anthropic's is limited to selected Pro and Max subscribers who use cloud sessions in Claude Code and don't already have projects on the web or desktop, with a waitlist if you don't have it yet. Anthropic's threads can't reach files on your machine or anything behind your network yet, though the post says that is coming very soon.

📰 News items and dev picks

AI news

Safety and governance

• We Must Pace the Frontier - Dario Amodei argued that capability gains are outrunning safety work, and proposed that frontier labs start by giving embedded third-party evaluators ongoing, employee-like access, then agree common safety standards between labs in democratic countries, then try to bring authoritarian governments into the same arrangement.
• Anthropic's threat intelligence report - Anthropic's report covers activity it disrupted between December 2025 and August 2026 across seven harm areas, including a cell of threat actors in northern Yemen using Claude for guided-rocket flight control firmware and multi-stage ballistic missile simulation, and likely freelance Russia-based actors using Claude Code to build a full-stack autonomous FPV drone swarm.
• Anthropic's own numbers - Anthropic proposed three transparency measurements for how fast a lab is moving and published its own figures against them, including that as of August 2026 Claude "leads" 26% of Anthropic's AI R&D work, up from under 1% in February.
• OpenAI's agents attacked RubyGems - Researchers say they believe internal OpenAI agents submitted over 2,000 malicious packages to RubyGems on May 11-12 and gained arbitrary remote code execution on RubyDoc.info's servers through user-specified .yardopts files, and their understanding from talking to people in the RubyGems community is that OpenAI never informed them it was responsible.
• 25 Fields Medallists on AI in maths - A declaration signed by 25 Fields Medal winners including Terence Tao argues that fast AI solutions published without proper write-ups undermine the point of the work and raise attribution and plagiarism problems, and the Clay Mathematics Institute said the same day that Navier-Stokes has "apparently been settled" but that its own review is "deliberately unhurried".

Coding agents and editors

• Claude Code 2.1.269 to 2.1.276 - Skills and plugins you have enabled on your claude.ai account now sync into terminal sessions, claude plugin eval runs plugin eval suites for scored and reproducible results, and auto mode with sandboxing gained per-command network allowlists for Bash, PowerShell and Monitor.
• Codex CLI 0.155.0 - Codex added experimental /voice conversations with live transcripts, live reasoning summaries in the status row, and Touch ID verification for MCP requests in local TUI sessions on supported Macs.
• VS Code 1.138 - Agents can now run inside a local Dev Container with your project's own tools and dependencies, and you can move the same Codex session between the ChatGPT app and VS Code.
• Copilot cost and quality dial - Copilot's automatic model selection is rolling out three tiers - efficiency, balance and intelligence - so you can choose how it weighs cost, quality and response time for each prompt.
• Xcode 27 - Xcode 27 reached general release on September 14 alongside iOS 27, iPadOS 27 and macOS 27, bringing models and agents from Anthropic, Google and OpenAI into the IDE, with interactive planning, multiturn Q&A, and tools that let coding agents validate their own work so they can run autonomously for longer.

New models and events

• Gemini 3.8 Live - Google released two live dialogue models to the Gemini API and AI Studio, and the Extended Thinking one reasons and speaks at the same time so it can narrate its way through a multi-step task.
• TypeSafe AI's Jev - Diogo Almeida, who co-invented RLHF and InstructGPT at OpenAI, launched a model that returns typed probabilistic decisions in a single parallel pass instead of generating text token by token, which he says means it cannot hallucinate or produce a type error.
• OpenAI DevDay - OpenAI DevDay is on September 29 at Fort Mason in San Francisco, with the keynote livestreamed for free and DevDay Exchanges following in eight more cities including London.

Dotnet news

• Performance Improvements in .NET 11 - Stephen Toub's annual deep dive into what got faster in .NET 11 is out.
• Rider and ReSharper 2026.2.2 - Rider gained an AI Agent Setup widget in the status bar that shows which coding-agent features are switched on, including MCP tools, skills and hooks, and dotCover can now measure coverage for TUnit tests.
• Intelligent Terminal 0.2.2572 - Microsoft's AI terminal added API-key support for local and OpenAI-compatible models, with the keys stored in Windows Credential Manager, an opt-in automatic approval mode for agents, and a 56% drop in median agent connection time for a single tab.
• Quill - Oren Eini, twenty years on from starting RavenDB, announced a layer that points at an existing Postgres, SQL Server, MariaDB or MySQL database and handles the synchronisation, embedding generation, vector search and agent orchestration on top of it.

Others

• Java 27 - JDK 27 reached general availability with nine JEPs, making G1 the default garbage collector in all environments, adding post-quantum hybrid key exchange for TLS 1.3, and turning on compact object headers by default.
• CUDA Rust - NVIDIA released two open-source projects for writing GPU kernels in Rust instead of C++, with cuda-oxide targeting the SIMT model on a pinned nightly toolchain and cutile-rs targeting the newer Tile model on stable Rust and installing from crates.io.
• Homebrew 7.0.0 - Homebrew 7 added a brew vulns vulnerability check backed by an advisory database and a native macOS app called BrewUI, and it drops macOS 10.15 and moves Intel Macs to tier 3 with no new bottles, ahead of Homebrew ceasing to run on Intel at all in September 2027.
• LiteLLM auth bypass - Wiz scanned around 3,000 internet-facing LiteLLM deployments and found nearly one in ten accepting a default master key or no authentication at all, then chained that to root-level code execution and cloud credential theft, and CISA added the MCP authentication bypass to its Known Exploited Vulnerabilities catalogue on September 2, though both flaws were patched months ago, in v1.82.0 and v1.84.0.
• Hacking OpenAI - Researchers chained a heap buffer overflow in libheif, reached through image uploads on OpenAI's Discourse forum, to an SSO misconfiguration that let them take over OpenAI employee accounts and reach connected services including GitHub, with OpenAI confirming a fix roughly 14 hours after the July 25 report and Discourse publishing its advisory three days later.

Sponsorship opportunities

I'm looking for sponsors for both the podcast and this newsletter. Details of podcast sponsorship can be found here. Feel free to reply to this email to discuss further.

The Podcast 🎙

A couple of weeks ago I recorded two episodes with Gui Ferreira. The first is about Claude Code, with plenty of tips and tricks along the way, and it should be out later today - my editing has run behind. The second is about building personal software, and using coding agents for things that aren't coding, and that one will follow a week or so later.

Gui was also on episode 78 last year, talking about goal setting, time management and productivity.

And if you missed the last one, it was Agent Skills with Matt Pocock.

Also, just a reminder that we have a Discord community for the podcast! It would be great to see you there :)

Podcast Discord community

My Dometrain Courses

Below are details of my Dometrain courses. You can also find all of my courses here.

Docker for Developers

This course will teach you everything you need to know about Docker and containers. From what containers and images are; to how to build your own; to security and networking; docker-compose; and much more!

From Zero to Hero: Docker for Developers

Kubernetes for Developers

Once you understand Docker, containers, images, etc - it's time to move onto the next level and learn a container orchestrator - and Kubernetes has clearly won the battle here! My Kubernetes course is rammed-packed full of demos (pretty much all the way through), which are easy to follow along with downloadable YAML files and scripts. We start with the basics, then later move on to more advanced topics like service meshes and operators.

From Zero to Hero: Kubernetes for Developers

JetBrains Rider

Rider is an amazing .NET IDE by JetBrains. This course is 6 hours of content - covering hotkeys, refactoring, navigation, debugging, git, testing, AI, profiling, remote collaboration, and much much more!

Aspire

Aspire is Microsoft's vision for how every .NET developer will develop systems. It is a cloud-ready stack for production-ready, distributed applications using .NET, and it makes it extremely easy to develop, run, and test your systems locally. With unmatched dev-time orchestration, integrations with third-party services, and excellent tooling, .NET Aspire aims to make working with the cloud and technologies like Docker and Kubernetes easier than ever.

Please help me share this newsletter 🙏

If you've made it this far into the newsletter - I'm hoping that means you've both enjoyed it and found it useful. If you can help me out and share with your developer friends at work and on social media, that would be amazing!

Again - feel totally free to reach out to me, and let me know your thoughts on the newsletter. And see you back in your inbox next week for the next edition! 👋

My socials...

X/Twitter (I've started using this again more proactively now)
Bluesky
LinkedIn
Discord
Mastodon (rarely used)

Or just go to danclarke.com for links to all my stuff 😊