<h2>Must know</h2>
<ul>
<li><strong>Microsoft patch tuesday addresses zero-days and AI vulnerabilities</strong>: Microsoft’s February 2026 Patch Tuesday is a big one, tackling 59 vulnerabilities, including six actively exploited zero-days. What’s particularly interesting for us is the patching of critical remote code execution flaws in GitHub Copilot and multiple IDEs (VS Code, Visual Studio, JetBrains), stemming from AI prompt injection vulnerabilities. This highlights the growing attack surface introduced by AI tools in our development environments, as reported by Krebs on Security. <a href="https://app.daily.dev/feed-by-ids?id=zk4Rm0PBv&amp;id=BOPPPXswR&amp;id=i2YjgAdJA&amp;id=oUCth2teJ&amp;id=uxvhPzT2r&amp;id=zFYu8Wov8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents introduce new security risks and governance challenges</strong>: The rapid deployment of autonomous AI agents, expected in 40% of enterprise apps by 2026, brings significant security vulnerabilities and policy concerns. GitGuardian, which just raised $50M, is focusing on securing these ‘Non-Human Identities’ and addressing gaps like prompt injection and supply chain attacks on AI agent ‘skills.’ This shift demands a move from identity-based to task-based authorization with just-in-time access, as highlighted by multiple security experts. <a href="https://app.daily.dev/feed-by-ids?id=0rQwEgnbB&amp;id=QRBfTNO1G&amp;id=ZVqULJLz3&amp;id=gUwIhEnZO&amp;id=l8YCDT7qf&amp;id=mmQEWnSC4&amp;id=1X0JlaofS&amp;id=48BFhb5Uw&amp;id=73Kh0XGFs&amp;id=8QVLeUBMy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open-source GLM-5 model claims parity with proprietary AI</strong>: <a href="http://Z.ai" target="_blank" rel="noopener nofollow">Z.ai</a> has released GLM-5, a new open-source large language model under an MIT License, claiming to match proprietary models like Claude Opus 4.6 and GPT 5.3 Codex for long-horizon coding tasks. With 744B parameters and novel asynchronous RL infrastructure, it’s a significant development in the open-source AI space, offering free access via Modal until April. <a href="https://app.daily.dev/posts/bOspGaOnW" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>ESLint 10.0.0 drops .eslintrc support for flat config</strong>: ESLint 10.0.0 is making flat config the only option, improving monorepo resolution and fixing JSX tracking for React projects, with no opt-out available. <a href="https://app.daily.dev/posts/eQ0JUyyzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP 29.0-rc1 introduces native records and quantum-resistant SSH</strong>: Erlang/OTP 29.0-rc1 brings native records, multi-valued comprehensions, and defaults SSH to quantum-resistant key exchange, while dropping 32-bit Windows builds. <a href="https://app.daily.dev/posts/fTCktIrzF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FastAPI cloud simplifies deployment with single command</strong>: FastAPI Cloud, backed by Sequoia, offers <code>fastapi deploy</code> for simplified FastAPI application deployment, handling HTTPS, auto-scaling, and Kubernetes infrastructure. <a href="https://app.daily.dev/feed-by-ids?id=5Mzv43k5F&amp;id=5atbRAfWV&amp;id=6Rrig1lfL&amp;id=7B5EDflI8&amp;id=N2p8QaBAl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches lakeflow for unified data engineering</strong>: Databricks introduced Lakeflow on Azure, a unified data engineering solution for ingestion, transformation, orchestration, and governance, featuring AI-powered debugging and serverless compute. <a href="https://app.daily.dev/feed-by-ids?id=Bf1dK14FP&amp;id=DckO36Vaq&amp;id=cW6KgKgEK&amp;id=omSTWjEfj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pandas 3.0.0 introduces copy-on-write and dedicated string dtype</strong>: Pandas 3.0.0 is out with breaking changes, including formal Copy-on-Write semantics to eliminate view/copy ambiguity and a dedicated string dtype, requiring Python 3.11+. <a href="https://app.daily.dev/feed-by-ids?id=Bf1dK14FP&amp;id=DckO36Vaq&amp;id=cW6KgKgEK&amp;id=omSTWjEfj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebMCP standard proposed for AI agent web interaction</strong>: A new WebMCP standard is proposed for Chrome 146 Canary, allowing websites to expose structured tools directly to in-browser AI agents, moving beyond brittle DOM scraping. <a href="https://app.daily.dev/feed-by-ids?id=Ib422mkez&amp;id=NpNZ22j1b&amp;id=R0k8YNGTx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 improves intel tsx and udp performance</strong>: Linux 7.0 changes default Intel TSX mode to ‘auto’ for performance and achieves a 12% UDP receive performance boost on 100Gbit NICs through manual inlining. <a href="https://app.daily.dev/feed-by-ids?id=feLrRFW3p&amp;id=mdQGXm2CW&amp;id=s1iSS5CBn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make you obsolete, unless you buy our tools</strong>: Apparently, AI models have reached an ‘inflection point’ and are now capable of autonomously completing complex, multi-hour tasks, eliminating the need for human oversight. Industry leaders are predicting 50% of entry-level white-collar jobs will vanish in 1-5 years, so you better immediately adopt paid AI tools and experiment daily, or face competitive disadvantage. Because, you know, the only way to survive the AI apocalypse is to buy more software. <a href="https://app.daily.dev/feed-by-ids?id=jQIyW7C3o&amp;id=lGyFb7XTB&amp;id=waJpUtc5V" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Replit’s AI app builder: build a real app in minutes, no coding required</strong>: Replit is showcasing a mobile app builder that lets you create, test, and publish apps using natural language prompts, claiming you can do it ‘within minutes’ and ‘without traditional coding.’ Because, clearly, the complex process of app development, debugging, and deployment can be magically reduced to a few sentences and a click. What could possibly go wrong when non-developers are churning out apps ‘without traditional coding’? <a href="https://app.daily.dev/feed-by-ids?id=XLMCxLn6R&amp;id=bMxG6ez54&amp;id=l8hipFss6&amp;id=niHDxajiA&amp;id=qWp2u9GdQ&amp;id=rSwW9H78J&amp;id=ryrr13JLg&amp;id=sH5JqveVi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The death of traditional testing, brought to you by AI</strong>: Datadog and GitHub Copilot are pushing new AI-driven testing solutions, with one article boldly proclaiming ‘The Death of Traditional Testing.’ Apparently, agentic development has ‘broken traditional testing paradigms,’ and now LLM-generated tests will ‘auto-heal’ themselves, eliminating maintenance costs. Because, of course, AI is so perfect it can write and maintain tests better than any human, and we’ll never need to manually verify anything again. What a time to be alive, where our tests write themselves and fix themselves, leaving us with absolutely nothing to do. <a href="https://app.daily.dev/feed-by-ids?id=gZJqHPom7&amp;id=j254S16VC&amp;id=sJ5P2kSW8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
