<h2>Must know</h2>
<ul>
<li><strong>AI coding agents are here, but still need adult supervision</strong>: AI coding agents like Claude Code are showing impressive velocity gains (10-100x) for well-defined tasks, even building a browser in Rust with over a million lines of code in a week, as reported by Cursor. However, they struggle with ambiguous requirements, often repeat code, and introduce vulnerabilities, according to research cited in J24eplQdr. Only a third of developers trust AI-generated code, highlighting the critical need for human oversight, strong test coverage, and architectural guardrails to prevent technical debt and quality regressions. <a href="https://app.daily.dev/feed-by-ids?id=EqMZfdLoq&amp;id=GPLZOgOgP&amp;id=J24eplQdr&amp;id=cSbwL1BS5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean hackers target fintech developers with VS Code exploits</strong>: North Korean threat actors are exploiting VS Code projects to target fintech, crypto, and blockchain developers. They distribute malicious job assessment projects that, when opened in VS Code, automatically execute payloads to deploy backdoors and malware, as detailed in technical reports. This campaign leverages developers’ trust in open-source workflows, enabling remote code execution and cryptocurrency mining on compromised machines. <a href="https://app.daily.dev/feed-by-ids?id=1wKLyk8gG&amp;id=Q9M3WuQfH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes is now the standard for enterprise AI, culture is the new bottleneck</strong>: Kubernetes production usage has hit 82%, with 66% of AI workloads using it for inference, according to the 2025 CNCF Annual Cloud Native Survey. This establishes Kubernetes as the standard for enterprise AI infrastructure. The survey also reveals that cultural resistance, not technical complexity, is now the primary barrier to adoption, indicating a shift in challenges for cloud-native and AI integration. <a href="https://app.daily.dev/posts/Lwsgn3lZ5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Next.js 16 boosts performance and AI debugging</strong>: Next.js 16 stabilizes Turbopack as the default bundler for faster builds, introduces explicit Cache Components, and integrates AI-assisted debugging via Model Context Protocol (MCP) in DevTools for AI agent-suggested fixes. <a href="https://app.daily.dev/posts/91xiEmcVe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic fixes critical Git server vulnerabilities</strong>: Anthropic patched three critical vulnerabilities in its Git MCP server (CVE-2025-68145, CVE-2025-68144, CVE-2025-68143) that could have allowed remote code execution via prompt injection, highlighting new attack surfaces in AI-integrated systems. <a href="https://app.daily.dev/posts/U6A8HrCeq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI accelerates exploit and malware development</strong>: LLMs like Opus 4.5 and GPT-5.2 generated over 40 zero-day exploits, bypassing modern security mitigations, and an AI-generated malware suite, VoidLink, was developed in under a week, demonstrating AI’s rapid acceleration of offensive cyber capabilities. <a href="https://app.daily.dev/feed-by-ids?id=WkT9zuzY1&amp;id=nLTC7GTyu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New EU decentralized vulnerability database launched</strong>: The GCVE has launched <a href="http://db.gcve.eu" target="_blank" rel="noopener nofollow">db.gcve.eu</a>, a public, decentralized cybersecurity vulnerability database integrating data from over 25 sources to reduce European reliance on US-based systems like CVE. <a href="https://app.daily.dev/posts/i2uCJ3RGE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft mandates Intune security updates for mobile apps</strong>: Microsoft will enforce mandatory Intune Mobile Application Management security updates from January 19, 2026, requiring iOS and Android apps to update to the latest SDK versions or be blocked from accessing corporate resources. <a href="https://app.daily.dev/posts/PNEoUht10" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebAssembly makes strides in browser and edge computing</strong>: WebAssembly saw significant progress in 2025 with Safari closing feature gaps, new proposals reaching standardization, and increased adoption, with WASI 0.3 and Component Model advancements expected in 2026. <a href="https://app.daily.dev/posts/jtf8mrjCJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WordPress 6.9.1 maintenance release planned</strong>: WordPress is planning a 6.9.1 maintenance release to address issues identified in support forums, Trac, and the Gutenberg repository following the 6.9 release. <a href="https://app.daily.dev/posts/hhrQQtQg3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Shadcn radio buttons: the pinnacle of over-engineering</strong>: Apparently, a simple HTML radio button, which has worked perfectly for 30 years, now requires ‘hundreds of lines of React code’ and ‘multiple dependencies’ from libraries like Shadcn and Radix. Because, you know, styling a native input with CSS is just too pedestrian. This isn’t innovation; it’s making a mountain out of a molehill, violating ARIA best practices, and adding unnecessary bloat, as pointed out by Lobsters. <a href="https://app.daily.dev/posts/cuZVekEgv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vibecoding: the new executive superpower (or just more AI hype)?</strong>: Emergent, an Indian AI startup, just bagged $70 million for ‘vibe-coding,’ claiming executives can now build full-stack apps without ‘advanced coding expertise.’ An opinion piece even suggests every exec should be ‘vibecoding.’ So, are we to believe that years of engineering discipline can be replaced by a ‘vibe’ and a $200 AI subscription? Sounds like another round of ‘AI will solve everything’ until the inevitable ‘AI-generated technical debt’ article drops. <a href="https://app.daily.dev/feed-by-ids?id=AllrxKqHW&amp;id=McILqauKu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated SwiftUI code: because memory leaks are a feature, not a bug</strong>: Who needs human engineers to write robust code when AI can generate SwiftUI that ‘persistently captures self strongly in deeply nested closures,’ leading to ‘unbounded memory growth’? The fix? Engineers had to ‘refactor code by adding weak references’ and ‘ownership annotations.’ It’s almost like AI isn’t quite ready to replace those pesky humans who actually understand memory management, as detailed in this postmortem. But hey, at least it was fast, right? <a href="https://app.daily.dev/feed-by-ids?id=0Dxh8ahUn&amp;id=NOlmy6yd2&amp;id=yTmR3XbQ9" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
