<h2>Must know</h2>
<ul>
<li><strong>React frameworks converge, offering developer choice</strong>: At React Conf, Kent Dodds and other framework creators noted that React frameworks are increasingly similar, giving developers freedom to choose without significant long-term risk. Next.js’s React Server Components, Expo Router’s file-based routing for React Native, and Redwood’s shift to a lightweight SDK were highlighted. All major frameworks are also exploring AI integration via Model Context Protocol (MCP) endpoints, as reported by The New Stack. <a href="https://app.daily.dev/feed-by-ids?id=Ii8I7s3X0&amp;id=oWHYd6BqQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firefox 147 and Safari TP 235 enhance web capabilities</strong>: Firefox 147 introduces automatic Picture-in-Picture, reorganized tab settings, and enhanced WebGPU support for Apple Silicon Macs, alongside stricter Enhanced Tracking Protection. Safari Technology Preview 235 adds support for name-only <code>@container</code> queries and the <code>math-depth</code> CSS property, improving layout and mathematical rendering capabilities. These updates, detailed by Frontend Focus and WebKit, continue to push modern web development forward. <a href="https://app.daily.dev/feed-by-ids?id=f7pzHnDhN&amp;id=rkOEUgQdM&amp;id=fzPCRf2as" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical ServiceNow vulnerability allows unauthenticated impersonation</strong>: AppOmni discovered ‘BodySnatcher’ (CVE-2025-12420), a critical ServiceNow flaw allowing unauthenticated attackers to impersonate users with just an email, bypassing MFA and SSO. This could lead to rogue AI agents with escalated privileges accessing external environments. ServiceNow has released a patch, but the incident highlights the growing attack surface with rapid AI agent adoption, according to AppOmni. <a href="https://app.daily.dev/posts/wzAqd15ET" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Livewire 4 introduces single-file components and performance boosts</strong>: Livewire 4 is out, bringing single-file components, an ‘islands’ feature for isolated rendering, native drag-and-drop, and View Transitions API integration, all while maintaining backward compatibility. <a href="https://app.daily.dev/posts/OxhgZzaoQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open-source PowerVR graphics drivers now available for BeaglePlay</strong>: The BeaglePlay board now supports fully open-source, upstream PowerVR graphics drivers, a significant shift addressing long-standing demands from the Linux and open-source communities. <a href="https://app.daily.dev/feed-by-ids?id=HwIiwOf9Q&amp;id=NjVG9GX8g" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CNCF graduates Dragonfly for cloud-native image distribution</strong>: Dragonfly, a peer-to-peer image and file distribution platform for container images and AI models, has graduated from CNCF, signaling its maturity and widespread adoption by major organizations like Alibaba and Datadog. <a href="https://app.daily.dev/posts/0rVfXzPjQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>U.S. agencies begin post-quantum security transition</strong>: U.S. federal agencies, including FBI, NIST, and CISA, are moving from awareness to active implementation of post-quantum security measures, releasing new cryptographic migration strategies and emphasizing immediate action against ‘harvest now, decrypt later’ threats. <a href="https://app.daily.dev/posts/lRZsyJgWP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CrowdStrike acquires Seraphic Security for browser threat protection</strong>: CrowdStrike is acquiring Seraphic Security for $400 million to integrate browser security into its Falcon platform, aiming to protect against session hijacking, malicious extensions, and data exfiltration via AI prompts. <a href="https://app.daily.dev/posts/AKJAUyPZD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Agoda centralizes financial data with Apache Spark</strong>: Agoda consolidated its financial data pipelines into a unified Apache Spark-based platform, FINUDP, to eliminate metric inconsistencies and process millions of booking transactions daily with hourly updates. <a href="https://app.daily.dev/feed-by-ids?id=SIYRILab7&amp;id=WuRILab7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MemOS launches as open-source memory OS for LLMs</strong>: MemOS, an open-source memory operating system for large language models and AI agents, launched on GitHub, offering unified long-term memory management for multi-modal data with claimed accuracy improvements and token savings. <a href="https://app.daily.dev/posts/bjqQdB8fJ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI solves math problems, but still needs a babysitter</strong>: OpenAI’s GPT 5.2 is apparently solving Erdős problems and ChatGPT is a ‘reasoning assistant’ in math research. But wait, Terence Tao says AI only excels at ‘obscure, straightforward’ problems and humans still need to verify proofs. So, AI can do the grunt work, but don’t expect it to invent new theorems or, you know, actually understand what it’s doing. It’s a calculator with a fancy autocomplete, not a genius. <a href="https://app.daily.dev/feed-by-ids?id=8o6H9wDXV&amp;id=mhyD4ij4A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel Boost adds AI prompts, because apparently we can’t simplify our own code anymore</strong>: Laravel Boost v1.8.10 now includes AI prompts for ‘code simplification’ and a Livewire v4 upgrade assistant. Because nothing says ‘efficient development workflows’ like outsourcing basic refactoring and upgrade tasks to an AI, especially one based on Claude. Are we really so busy exploring ‘modern web technologies’ that we can’t even clean up our own Laravel code without a bot holding our hand? <a href="https://app.daily.dev/posts/TQr7IJii0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Cowork: more AI, more ways to shoot yourself in the foot</strong>: Anthropic released Cowork, a new feature for Claude Code that lets it automate enterprise tasks and interact with databases and browsers. They highlight its ability to summarize calendars and create reports, but also casually mention ‘susceptibility to destructive commands from misunderstood prompts and prompt injection attacks.’ So, now your AI can not only hallucinate, but also delete your data if you phrase a prompt wrong. Progress! <a href="https://app.daily.dev/feed-by-ids?id=GMZHoSF60&amp;id=QvZHdydfi&amp;id=U2HzInRUP&amp;id=ejPwSyHLL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
