<h2>Must know</h2>
<ul>
<li><strong>AI is eating open-source revenue models</strong>: Tailwind CSS reported a 75% staff layoff and an 80% revenue drop, directly attributing it to AI models providing direct answers and code, thus bypassing their documentation-based monetization. This highlights a critical challenge for open-source projects that rely on documentation traffic for revenue, as AI consumes content without contributing back, as reported by Tailwind CSS itself. The company declined GitHub user requests to make their documentation more AI-friendly, citing adverse business impacts. <a href="https://app.daily.dev/feed-by-ids?id=RuMhQtj6W&amp;id=f3fYMXuKZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding agents are a double-edged sword</strong>: A software consultancy found that while AI tools speed up early ideation and simple code changes, they also introduce ‘anchoring effects’ that limit innovative thinking and create review bottlenecks. Over-reliance on AI can lead to ‘knowledge erosion’ among developers, making software harder to maintain, as detailed in a recent report. SonarSource also highlights that 96% of developers distrust AI-generated code, yet nearly half don’t always verify it before committing, leading to potential security vulnerabilities like SQL injection. <a href="https://app.daily.dev/feed-by-ids?id=DRTczTJyG&amp;id=K8Q8uNU3E&amp;id=Kf8FXuyo&amp;id=E02dc9OJZ&amp;id=q2sjuwRsf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean hackers weaponize QR codes for credential theft</strong>: The FBI warns that North Korean state-sponsored hackers (Kimsuky/APT43) are using malicious QR codes in spear-phishing campaigns targeting U.S. institutions. These ‘quishing’ attacks embed QR codes in emails that redirect users to credential harvesting sites, bypassing traditional email security and exploiting personal mobile devices. This technique allows attackers to steal session tokens and bypass multi-factor authentication, as reported by the FBI. <a href="https://app.daily.dev/posts/XmPhFlPf5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>VS Code introduces agent skills for AI agents</strong>: Visual Studio Code has launched an experimental ‘Agent Skills’ feature, allowing AI agents like GitHub Copilot to load specialized instructions and resources on demand for domain-specific developer tasks, enhancing automation and context-aware workflows. <a href="https://app.daily.dev/posts/xn90KJxf1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cursor AI editor challenges VS Code dominance</strong>: Cursor, an AI-native code editor, has gained 7 million users by deeply integrating AI, offering features like ‘Composer’ for autonomous code refactoring, though it faces challenges like Microsoft restricting its VS Code marketplace access and infrastructure instability. <a href="https://app.daily.dev/feed-by-ids?id=LbaA6unww&amp;id=qjJ5kf4y2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic updates Claude Code with async sub-agents and session teleportation</strong>: Anthropic released Claude Code 2.1, introducing forked and asynchronous sub-agents for parallel task execution, automatic skill hot reloading, and ‘session teleportation’ for seamless transitions between terminal and desktop environments. <a href="https://app.daily.dev/feed-by-ids?id=WChBmOscl&amp;id=nDBQDaZMg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google redesigns Gmail with AI-powered features</strong>: Google is launching new AI-powered features for Gmail, including AI Overviews to summarize email threads, natural language inbox queries, and enhanced writing assistance, all powered by Gemini 3, with advanced features requiring paid subscriptions. <a href="https://app.daily.dev/posts/8lOS5k1vH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Snowflake acquires Observe for AIops observability</strong>: Snowflake is acquiring Observe, an AI-driven SRE platform, to enhance observability for production AI systems by providing integrated visibility across data pipelines, AI model behavior, and infrastructure health, potentially making observability more cost-effective than traditional tools. <a href="https://app.daily.dev/posts/XlAd95QPv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New developments in Kubernetes database management and cloud databases</strong>: Percona is donating its database management platform, now OpenEverest, to the CNCF for unified provisioning and management of PostgreSQL, MySQL, and MongoDB on Kubernetes, while Amazon Lightsail introduced larger managed database bundles for production workloads. <a href="https://app.daily.dev/feed-by-ids?id=MhqrMxmeE&amp;id=waU9VD2zd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel Boost updates for Model Context Protocol (MCP)</strong>: Laravel Boost v1.8.9 now supports the Model Context Protocol (MCP), a framework enabling AI agents to access tools and data efficiently, with security experts emphasizing the need for Zero Trust principles and unique credentials for each AI agent. <a href="https://app.daily.dev/feed-by-ids?id=8827svTqM&amp;id=lqqfnNpcw" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make you a 10x engineer, just ignore the infrastructure instability</strong>: Apparently, Cursor, the AI-native code editor, is so revolutionary it ‘destroyed VSCode’ and helped 7 million users become coding gods. Never mind the ‘infrastructure instability’ and Microsoft restricting its marketplace access, which conveniently drove some developers back to VSCode. It seems the path to 10x productivity is paved with hype and a healthy dose of ‘just ignore the bugs,’ as long as the AI is ‘deeply integrated.’ <a href="https://app.daily.dev/feed-by-ids?id=LbaA6unww&amp;id=qjJ5kf4y2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are the future, just don’t forget to babysit them</strong>: The new Model Context Protocol (MCP) is here to enable AI agents to access tools and data ‘efficiently.’ But wait, security experts are already screaming about the need for Zero Trust, unique credentials, and granular authorization because these agents ‘lack human judgment’ and will try ‘all possible accessible actions.’ So, we’re building autonomous agents that need constant human oversight to prevent them from doing something stupid. Progress! <a href="https://app.daily.dev/feed-by-ids?id=8827svTqM&amp;id=lqqfnNpcw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Generative AI: the ‘cancerous’ threat to game development, unless it’s Larian Studios doing it</strong>: Hooded Horse, the publisher behind Manor Lords, is calling generative AI ‘cancerous’ and contractually banning it from all its partnered studios, citing copyright concerns and the risk of AI-created assets ‘slipping’ into games. Meanwhile, Larian Studios, after some ‘public disapproval,’ has clarified they won’t use generative AI for Divinity concept art, but will ‘continue internal research’ and restrict use to models trained on ‘proprietary data’ for in-game assets. So, it’s ‘cancerous’ for the little guys, but for the big studios, it’s just ‘internal research’ and ‘proprietary data.’ Got it. <a href="https://app.daily.dev/feed-by-ids?id=ufoxpOMSz&amp;id=cBuEU5Zrv&amp;id=cDdqzyRVb&amp;id=o9kOEf5qk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
