<h2>Must know</h2>
<ul>
<li><strong>Next.js 16.2 and Cursor Composer 2 accelerate AI native dev</strong>: Next.js 16.2 introduces a dedicated <a href="http://agents.md" target="_blank" rel="noopener nofollow">agents.md</a> file for AI context and experimental dev tools that expose browser-level data like component trees directly to coding agents. Simultaneously, Cursor released Composer 2, an in-house model using ‘self-summarization’ to compress context windows to 1,000 tokens, which reportedly outperforms Claude 4.6 on coding benchmarks. These releases signal a shift from AI as a sidecar to AI as a primary architectural citizen in the development loop. <a href="https://app.daily.dev/feed-by-ids?id=pLH5NluEB&amp;id=mIKxoNmwG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Interlock ransomware exploits Cisco zero day via deserialization</strong>: The Interlock ransomware group exploited a critical zero-day vulnerability (CVE-2026-20131) in Cisco Secure Firewall Management Center for 38 days before a patch was issued. Amazon’s MadPot honeypot network identified the attack chain, which allows remote code execution with a maximum CVSS score of 10. Security teams are urged by Cisco and Amazon to patch immediately to prevent further exploitation by this RaaS group. <a href="https://app.daily.dev/posts/kS1FWAP6y" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FBI seizes Handala domains following Stryker infrastructure wipe</strong>: The FBI and DOJ seized websites belonging to the pro-Iranian group Handala after a devastating attack on medical tech giant Stryker. The group compromised an admin account to hijack Microsoft Intune dashboards, remotely wiping tens of thousands of employee devices and knocking supply systems offline. CISA has issued an urgent alert via its security portal recommending dual-administrator approval for high-impact Intune actions like device wipes. <a href="https://app.daily.dev/posts/N8kFe3mft" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Heroku CLI v11 completes ESM migration and doubles slug limits</strong>: The overhaul moves to Node 22 and oclif v4, increasing slug limits to 1GB to better accommodate heavy AI and machine learning frameworks. <a href="https://app.daily.dev/feed-by-ids?id=2swWgPS9c&amp;id=vu5ItVeI2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TanStack Start gains traction as a lightweight Next.js alternative</strong>: Developers are reporting significant velocity gains by migrating to TanStack Start (React 19 + Vite) to avoid the perceived over-engineering of the Next.js App Router. <a href="https://app.daily.dev/posts/4nwi9O4T1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches serverless NVIDIA H100 training runtime</strong>: The new AI Runtime (AIR) provides on-demand distributed GPU clusters with pre-installed PyTorch and Ray, integrated directly into the Lakehouse for LLM fine-tuning. <a href="https://app.daily.dev/posts/ofgaFNNzJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub adds docked panels for side by side PR reviews</strong>: A new UI update allows developers to view comments, merge status, and security alerts in persistent side panels while scrolling through code diffs. <a href="https://app.daily.dev/posts/g5KSNDYIS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS SDK for Ruby adds native S3 directory sync</strong>: The updated Transfer Manager now includes automated multipart handling and parallelism for bulk directory uploads and downloads. <a href="https://app.daily.dev/posts/tlRI3NVwm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Reloader controller automates Kubernetes rolling upgrades on config change</strong>: This open-source tool watches ConfigMaps and Secrets to trigger pod restarts, supporting GitOps workflows and external secret providers like Vault. <a href="https://app.daily.dev/posts/fOD5l8pkQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Avalonia MAUI brings .NET apps to Linux and WebAssembly</strong>: The first preview allows .NET MAUI applications to run on Linux by mapping MAUI abstractions to Avalonia controls without code changes. <a href="https://app.daily.dev/posts/wOaYwcJqb" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Google AI Studio and the vibe coding rebrand</strong>: Google is trying to make ‘vibe coding’ happen, a marketing term for what is essentially just a natural language wrapper around Cloud Run and Firebase. While the integration is slick, framing basic app generation as a ‘new era of development’ ignores the inevitable technical debt and architectural mess these ‘vibes’ create. It’s a classic attempt to rebrand prompt engineering for people who don’t want to learn how a backend actually works. <a href="https://app.daily.dev/feed-by-ids?id=vDejXlEPp&amp;id=TarXVudPr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LangSmith Fleet and the enterprise agent sprawl</strong>: LangSmith rebranded its agent builder to ‘Fleet,’ pitching it as a way to manage an ‘organization-wide fleet’ of AI agents. The announcement is heavy on ‘human-in-the-loop’ buzzwords and ‘Claws’ credential models, but light on how this actually solves the underlying reliability issues of agentic workflows. It feels like a solution searching for a problem, assuming enterprises are already drowning in agents rather than just struggling to get one to work reliably. <a href="https://app.daily.dev/posts/fnry81udE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Checkmarx and the AI security arms race</strong>: Checkmarx is leaning hard into ‘AI orchestration’ with a framework that manages five different AI agents to automate vulnerability triage. Calling it an ‘AI arms race’ that requires ‘machine-speed security’ is peak marketing fear-mongering designed to sell more licenses. In reality, it’s just adding LLM-based heuristics to existing SAST/DAST tools and hoping the hallucinations don’t bury developers in even more false positives. <a href="https://app.daily.dev/posts/PE8tBh7Qo" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
