<h2>Must know</h2>
<ul>
<li><strong>AWS CodeBuild regex misconfiguration exposed supply chain to risk</strong>: Security researchers at Wiz discovered a critical vulnerability in AWS CodeBuild where improperly anchored regex patterns in webhook filters could have allowed unauthenticated attackers to bypass security controls and gain admin access to AWS GitHub repositories, including the JavaScript SDK. AWS patched the issue within 48 hours, but the incident highlights the persistent risks in CI/CD pipeline misconfigurations, as reported by CSO Online and Hacker News. This could have led to widespread supply chain attacks against all AWS environments. <a href="https://app.daily.dev/feed-by-ids?id=q7gLeN4EH&amp;id=hgS5v4ArI&amp;id=MZnlHztN0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mandiant publishes NTLMv1 rainbow tables to force deprecation</strong>: Mandiant, a Google-owned cybersecurity firm, released comprehensive rainbow tables for Net-NTLMv1 authentication, making it trivial to crack credentials in under 12 hours with consumer-grade hardware. This move aims to force organizations to abandon the 20-year-old insecure protocol, still prevalent in legacy systems, despite Microsoft removing its support in Windows Server 2025. The tables are publicly hosted on Google Cloud, increasing accessibility for both defenders and attackers, as detailed across multiple security news outlets. <a href="https://app.daily.dev/posts/2q2phjvWS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical Fortinet FortiSIEM RCE vulnerability disclosed</strong>: Fortinet patched CVE-2025-64155, a critical remote code execution vulnerability in FortiSIEM (versions 6.7 through 7.4.0) that allows unauthenticated attackers to achieve full system takeover via the phMonitor service. A public proof-of-concept exploit is available, increasing the urgency for immediate upgrades to patched versions (7.1.9+, 7.2.7+, 7.3.5+, and 7.4.1+). This flaw is part of a recurring pattern of root access vulnerabilities in the same FortiSIEM component, often exploited by nation-state actors. <a href="https://app.daily.dev/posts/2ys9nRyqb" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Cilium operational best practices for large Kubernetes clusters</strong>: Datadog shares key operational practices for running Cilium at enterprise scale, including using native routing, tuning IPAM for surge allocation, implementing strict upgrade gates, monitoring BPF map pressure, and configuring kube-proxy replacement with Maglev hashing. <a href="https://app.daily.dev/posts/QzUGYBGCW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>VoidLink malware targets Linux cloud servers</strong>: Check Point researchers discovered VoidLink, a sophisticated modular malware framework written in Zig that targets Linux cloud servers, adapting its behavior to major cloud platforms and containerized environments like Docker and Kubernetes. <a href="https://app.daily.dev/posts/TaWMZZCV9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pulumi IAM for self-hosted cloud and zero trust</strong>: Pulumi IAM now offers custom roles and scoped access tokens for self-hosted Pulumi Cloud instances, enabling fine-grained permissions and Zero Trust security for CI/CD automation. <a href="https://app.daily.dev/posts/wd03nDp6A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Dragonfly graduates CNCF for image and file distribution</strong>: Dragonfly, an open-source peer-to-peer file distribution system, has graduated from CNCF, enabling rapid distribution of container images, OCI artifacts, and AI models across thousands of nodes, saving up to 90% bandwidth and reducing launch times. <a href="https://app.daily.dev/feed-by-ids?id=0rVfXzPjQ&amp;id=z5gMS2kCc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>eBPF.party for interactive eBPF learning</strong>: eBPF.party is an interactive, open-source platform that allows users to write, compile, and run eBPF programs directly in their browser through hands-on exercises. <a href="https://app.daily.dev/posts/SBDx7e0x2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes operators with Go and Kubebuilder</strong>: A 6-hour video course teaches how to build custom Kubernetes operators and controllers from scratch using Go and Kubebuilder, covering controller theory, API building, and advanced internals like Informers and Finalizers. <a href="https://app.daily.dev/posts/lM0ntmppN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes as the unified control plane for VMs and containers</strong>: Industry predictions for 2026 suggest Kubernetes will emerge as the unified control plane for both VMs and containers, especially as an alternative to VMware, driven by cost pressures and hybrid infrastructure portability. <a href="https://app.daily.dev/posts/zVJaiO8vu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI coding agents: the new bottleneck is you</strong>: AI coding agents are supposedly reducing implementation time, shifting the bottleneck from execution to ‘problem definition and requirements gathering.’ So, instead of writing bad code, engineers now need to be ‘Product Engineers’ who can perfectly articulate what the AI should do. Because, apparently, the AI can’t figure out what the customer wants, but it can write perfect code if you just tell it exactly what to do. Sounds like the same old problem, just with more steps and a fancier title. <a href="https://app.daily.dev/feed-by-ids?id=ruSCTD0Fh&amp;id=sifgyKGyd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered security: because your employees are the real threat</strong>: Living Security is beta testing ‘Livvy,’ an AI engine that analyzes ‘billions of signals’ to predict user risk and identify ‘risky end-user behaviors.’ Because nothing says ‘trust’ like an AI constantly monitoring your every move, especially when ‘AI agents proliferate and outpace the capabilities of traditional cybersecurity oversight.’ So, instead of fixing the AI’s security flaws, we’re just going to surveil the humans more intensely. Problem solved, right? <a href="https://app.daily.dev/feed-by-ids?id=c8Xk5dH7c&amp;id=gKUBAdf71" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Small language models: the new ‘disruptive innovation’ that’s actually just cheaper</strong>: Enterprises are suddenly ‘favoring’ Small Language Models (SLMs) over LLMs because of ‘cost, latency, and fit for closed, well-defined workflows.’ Apparently, LLMs are now a ‘liability’ because they’re too expensive and open-ended. It’s almost as if the industry realized that throwing a 100-billion-parameter behemoth at every problem isn’t always the most efficient solution. Who knew that smaller, specialized tools could be better for specific tasks? Revolutionary. <a href="https://app.daily.dev/feed-by-ids?id=HX43N3MkE&amp;id=J1BKrq2u6" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
