<h2>Must know</h2>
<ul>
<li><strong>OpenAI acquires Astral to dominate Python infrastructure</strong>: OpenAI has acquired Astral, the team behind the high-performance Python tools uv and ruff, which currently see over 126 million monthly downloads. While both parties promise continued open-source support, analysts like Simon Willison suggest this is a strategic talent grab for the Codex team to counter Anthropic’s acquisition of the Bun ecosystem. The move raises concerns about OpenAI potentially leveraging ownership of critical infrastructure against competitors or shifting focus away from Astral’s private registry products. <a href="https://app.daily.dev/posts/QlXSsg9ai" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical OpenWRT XSS grants unauthenticated root access</strong>: A severe vulnerability (CVE-2026-32721) in the LuCI web interface allows attackers to gain persistent root SSH access simply by broadcasting a malicious WiFi SSID. By exploiting unsanitized innerHTML in the wireless scan page, an attacker can inject JavaScript to plant an SSH public key via the ubus API without any user credentials. Fixes are now available in OpenWrt 24.10.6 and 25.12.1; administrators should update immediately before performing any wireless scans. <a href="https://app.daily.dev/posts/YnILg1RdD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft unifies SQL and Fabric into an AI-native data platform</strong>: At SQLCon 2026, Microsoft announced the convergence of its database portfolio and Microsoft Fabric into a single platform featuring deep GitHub Copilot integration for T-SQL. New capabilities include a SQL MCP Server for connecting AI agents to Azure SQL Hyperscale and a Database Hub for unified management across Cosmos DB, PostgreSQL, and MySQL. The update also introduces OneLake mirroring for Snowflake and Databricks, positioning Fabric as a governed foundation for enterprise ‘Agentic AI’ workflows. <a href="https://app.daily.dev/posts/9lGZhe2Qp" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Next.js 16.2 boosts performance and adds AI agent context</strong>: The release features 400% faster dev server starts and a new <a href="http://agents.md" target="_blank" rel="noopener nofollow">agents.md</a> file to provide context for AI coding tools by default. <a href="https://app.daily.dev/posts/pLH5NluEB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Sashiko brings AI-powered code review to the Linux kernel</strong>: Developed by Google and now under the Linux Foundation, this Rust-based tool identified 53% of bugs in testing that human reviewers missed. <a href="https://app.daily.dev/posts/j5mdnOwUC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PostgreSQL 19 introduces built-in REPACK command</strong>: The new command consolidates VACUUM FULL and CLUSTER functionality to reclaim disk space from table bloat without manual extensions. <a href="https://app.daily.dev/posts/NAfFQZLbk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Monty is a Rust-based Python interpreter for safe AI agents</strong>: Created by the Pydantic team, Monty offers 6µs cold starts and a fully sandboxed environment designed specifically for executing LLM-generated code. <a href="https://app.daily.dev/posts/ZrXDJ2QU8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PowerShell 7.6 LTS lands on .NET 10</strong>: The new long-term support release includes dozens of tab completion enhancements and promoted experimental features for production automation. <a href="https://app.daily.dev/posts/TIHXwNFB4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Modern CSS is killing off heavy JavaScript libraries</strong>: Native APIs for Anchor Positioning, Popover, and Scroll-Driven Animations can now replace up to 322kB of minified JS from libraries like GSAP and Popper. <a href="https://app.daily.dev/posts/OTVoRLt9s" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>China to diverge from NIST with structureless lattice PQC standards</strong>: Chinese researchers plan to establish independent post-quantum cryptography standards within three years, citing potential security weaknesses in NIST’s algebraic lattice designs. <a href="https://app.daily.dev/posts/oyhepMDNM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Vibe coding is just marketing for ‘I hope this prompt works’</strong>: Google is pushing ‘vibe coding’ in AI Studio as a revolutionary shift to AI-native development, but it’s essentially just a high-level wrapper for generating boilerplate and deploying to Cloud Run. While Cisco execs claim it turned a year-long project into a week of work, they conveniently omit the ‘extensive validation’ required to fix the inevitable hallucinations. It’s a rebranding of low-code tools for the LLM era, designed to make non-technical managers feel like software architects. <a href="https://app.daily.dev/feed-by-ids?id=vDejXlEPp&amp;id=0hn0P8EbP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Corporate AI mandates are the new surveillance state</strong>: Amazon and PwC are now tying AI tool usage directly to performance reviews and job security, with Amazon’s ‘Clarity’ system tracking developer activity weekly. This forced adoption ignores research showing a 17% drop in skill retention and the fact that half of businesses see zero benefit from these tools. It’s a classic management pivot: if you can’t prove productivity gains, just mandate the metrics until the dashboard looks green. <a href="https://app.daily.dev/feed-by-ids?id=9KgLLeZB1&amp;id=BbYdrSqWL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The 1000x faster database that forgot about the network</strong>: SpacetimeDB is making the rounds claiming it’s 1000x faster than Postgres by ‘co-locating’ code and data in RAM. This isn’t a breakthrough; it’s just rediscovering that in-memory computing is faster than disk I/O if you ignore the complexities of distributed state and network latency. Calling it a Postgres killer is like saying a calculator is a faster supercomputer because it doesn’t have to wait for a login shell. <a href="https://app.daily.dev/posts/fgMT5xF5i" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
